Groundwork for monitoring and updating the dns

Former-commit-id: baf38fac2f

.env.example

@@ -0,0 +1,5 @@
+# Copy this file to .env and fill in your values
+# Required for connecting to Pangolin service
+PANGOLIN_ENDPOINT=https://app.pangolin.net
+NEWT_ID=changeme-id
+NEWT_SECRET=changeme-secret

.github/CODEOWNERS

@@ -0,0 +1 @@
+* @oschwartz10612 @miloschwartz

.github/DISCUSSION_TEMPLATE/feature-requests.yml

@@ -0,0 +1,47 @@
+body:
+    - type: textarea
+      attributes:
+          label: Summary
+          description: A clear and concise summary of the requested feature.
+      validations:
+          required: true
+
+    - type: textarea
+      attributes:
+          label: Motivation
+          description: |
+              Why is this feature important?
+              Explain the problem this feature would solve or what use case it would enable.
+      validations:
+          required: true
+
+    - type: textarea
+      attributes:
+          label: Proposed Solution
+          description: |
+              How would you like to see this feature implemented?
+              Provide as much detail as possible about the desired behavior, configuration, or changes.
+      validations:
+          required: true
+
+    - type: textarea
+      attributes:
+          label: Alternatives Considered
+          description: Describe any alternative solutions or workarounds you've thought about.
+      validations:
+          required: false
+
+    - type: textarea
+      attributes:
+          label: Additional Context
+          description: Add any other context, mockups, or screenshots about the feature request here.
+      validations:
+          required: false
+
+    - type: markdown
+      attributes:
+          value: |
+              Before submitting, please:
+              - Check if there is an existing issue for this feature.
+              - Clearly explain the benefit and use case.
+              - Be as specific as possible to help contributors evaluate and implement.

.github/ISSUE_TEMPLATE/1.bug_report.yml

@@ -0,0 +1,52 @@
+name: Bug Report
+description: Create a bug report
+labels: []
+body:
+    - type: textarea
+      attributes:
+          label: Describe the Bug
+          description: A clear and concise description of what the bug is.
+      validations:
+          required: true
+
+    - type: textarea
+      attributes:
+          label: Environment
+          description: Please fill out the relevant details below for your environment.
+          value: |
+              - OS Type & Version:
+              - Pangolin Version:
+              - Edition (Community or Enterprise):
+              - Gerbil Version:
+              - Traefik Version:
+              - Newt Version:
+              - Client Version:
+      validations:
+          required: true
+
+    - type: textarea
+      attributes:
+          label: To Reproduce
+          description: |
+              Steps to reproduce the behavior, please provide a clear description of how to reproduce the issue, based on the linked minimal reproduction. Screenshots can be provided in the issue body below.
+
+              If using code blocks, make sure syntax highlighting is correct and double-check that the rendered preview is not broken.
+      validations:
+          required: true
+
+    - type: textarea
+      attributes:
+          label: Expected Behavior
+          description: A clear and concise description of what you expected to happen.
+      validations:
+          required: true
+
+    - type: markdown
+      attributes:
+          value: |
+              Before posting the issue go through the steps you've written down to make sure the steps provided are detailed and clear.
+
+    - type: markdown
+      attributes:
+          value: |
+              Contributors should be able to follow the steps provided in order to reproduce the bug.

.github/ISSUE_TEMPLATE/config.yml

@@ -0,0 +1,8 @@
+blank_issues_enabled: false
+contact_links:
+    - name: Need help or have questions?
+      url: https://github.com/orgs/fosrl/discussions
+      about: Ask questions, get help, and discuss with other community members
+    - name: Request a Feature
+      url: https://github.com/orgs/fosrl/discussions/new?category=feature-requests
+      about: Feature requests should be opened as discussions so others can upvote and comment

.github/dependabot.yml

@@ -0,0 +1,40 @@
+version: 2
+updates:
+  - package-ecosystem: "gomod"
+    directory: "/"
+    schedule:
+      interval: "daily"
+    groups:
+      dev-patch-updates:
+        dependency-type: "development"
+        update-types:
+          - "patch"
+      dev-minor-updates:
+        dependency-type: "development"
+        update-types:
+          - "minor"
+      prod-patch-updates:
+        dependency-type: "production"
+        update-types:
+          - "patch"
+      prod-minor-updates:
+        dependency-type: "production"
+        update-types:
+          - "minor"
+
+  - package-ecosystem: "docker"
+    directory: "/"
+    schedule:
+      interval: "daily"
+    groups:
+      patch-updates:
+        update-types:
+          - "patch"
+      minor-updates:
+        update-types:
+          - "minor"
+
+  - package-ecosystem: "github-actions"
+    directory: "/"
+    schedule:
+      interval: "weekly"

.github/workflows/mirror.yaml

@@ -0,0 +1,132 @@
+name: Mirror & Sign (Docker Hub to GHCR)
+
+on:
+  workflow_dispatch: {}
+
+permissions:
+  contents: read
+  packages: write
+  id-token: write   # for keyless OIDC
+
+env:
+  SOURCE_IMAGE: docker.io/fosrl/newt
+  DEST_IMAGE: ghcr.io/${{ github.repository_owner }}/${{ github.event.repository.name }}
+
+jobs:
+  mirror-and-dual-sign:
+    runs-on: amd64-runner
+    steps:
+      - name: Install skopeo + jq
+        run: |
+          sudo apt-get update -y
+          sudo apt-get install -y skopeo jq
+          skopeo --version
+
+      - name: Install cosign
+        uses: sigstore/cosign-installer@6f9f17788090df1f26f669e9d70d6ae9567deba6 # v4.1.2
+
+      - name: Input check
+        run: |
+          test -n "${SOURCE_IMAGE}" || (echo "SOURCE_IMAGE is empty" && exit 1)
+          echo "Source : ${SOURCE_IMAGE}"
+          echo "Target : ${DEST_IMAGE}"
+
+      # Auth for skopeo (containers-auth)
+      - name: Skopeo login to GHCR
+        run: |
+          skopeo login ghcr.io -u "${{ github.actor }}" -p "${{ secrets.GITHUB_TOKEN }}"
+
+      # Auth for cosign (docker-config)
+      - name: Docker login to GHCR (for cosign)
+        run: |
+          echo "${{ secrets.GITHUB_TOKEN }}" | docker login ghcr.io -u "${{ github.actor }}" --password-stdin
+
+      - name: List source tags
+        run: |
+          set -euo pipefail
+          skopeo list-tags --retry-times 3 docker://"${SOURCE_IMAGE}" \
+            | jq -r '.Tags[]' | sort -u > src-tags.txt
+          echo "Found source tags: $(wc -l < src-tags.txt)"
+          head -n 20 src-tags.txt || true
+
+      - name: List destination tags (skip existing)
+        run: |
+          set -euo pipefail
+          if skopeo list-tags --retry-times 3 docker://"${DEST_IMAGE}" >/tmp/dst.json 2>/dev/null; then
+            jq -r '.Tags[]' /tmp/dst.json | sort -u > dst-tags.txt
+          else
+            : > dst-tags.txt
+          fi
+          echo "Existing destination tags: $(wc -l < dst-tags.txt)"
+
+      - name: Mirror, dual-sign, and verify
+        env:
+          # keyless
+          COSIGN_YES: "true"
+          # key-based
+          COSIGN_PRIVATE_KEY: ${{ secrets.COSIGN_PRIVATE_KEY }}
+          COSIGN_PASSWORD:    ${{ secrets.COSIGN_PASSWORD }}
+          # verify
+          COSIGN_PUBLIC_KEY:  ${{ secrets.COSIGN_PUBLIC_KEY }}
+        run: |
+          set -euo pipefail
+          copied=0; skipped=0; v_ok=0; errs=0
+
+          issuer="https://token.actions.githubusercontent.com"
+          id_regex="^https://github.com/${{ github.repository }}/.+"
+
+          while read -r tag; do
+            [ -z "$tag" ] && continue
+
+            if grep -Fxq "$tag" dst-tags.txt; then
+              echo "::notice ::Skip (exists) ${DEST_IMAGE}:${tag}"
+              skipped=$((skipped+1))
+              continue
+            fi
+
+            echo "==> Copy ${SOURCE_IMAGE}:${tag} → ${DEST_IMAGE}:${tag}"
+            if ! skopeo copy --all --retry-times 3 \
+                 docker://"${SOURCE_IMAGE}:${tag}" docker://"${DEST_IMAGE}:${tag}"; then
+              echo "::warning title=Copy failed::${SOURCE_IMAGE}:${tag}"
+              errs=$((errs+1)); continue
+            fi
+            copied=$((copied+1))
+
+            digest="$(skopeo inspect --retry-times 3 docker://"${DEST_IMAGE}:${tag}" | jq -r '.Digest')"
+            ref="${DEST_IMAGE}@${digest}"
+
+            echo "==> cosign sign (keyless) --recursive ${ref}"
+            if ! cosign sign --recursive "${ref}"; then
+              echo "::warning title=Keyless sign failed::${ref}"
+              errs=$((errs+1))
+            fi
+
+            echo "==> cosign sign (key) --recursive ${ref}"
+            if ! cosign sign --key env://COSIGN_PRIVATE_KEY --recursive "${ref}"; then
+              echo "::warning title=Key sign failed::${ref}"
+              errs=$((errs+1))
+            fi
+
+            echo "==> cosign verify (public key) ${ref}"
+            if ! cosign verify --key env://COSIGN_PUBLIC_KEY "${ref}" -o text; then
+              echo "::warning title=Verify(pubkey) failed::${ref}"
+              errs=$((errs+1))
+            fi
+
+            echo "==> cosign verify (keyless policy) ${ref}"
+            if ! cosign verify \
+                 --certificate-oidc-issuer "${issuer}" \
+                 --certificate-identity-regexp "${id_regex}" \
+                 "${ref}" -o text; then
+              echo "::warning title=Verify(keyless) failed::${ref}"
+              errs=$((errs+1))
+            else
+              v_ok=$((v_ok+1))
+            fi
+          done < src-tags.txt
+
+          echo "---- Summary ----"
+          echo "Copied        : $copied"
+          echo "Skipped       : $skipped"
+          echo "Verified OK   : $v_ok"
+          echo "Errors        : $errs"

.github/workflows/nix-build.yml

@@ -0,0 +1,23 @@
+name: Build Nix package
+
+on:
+  workflow_dispatch:
+  pull_request:
+    paths:
+      - go.mod
+      - go.sum
+
+jobs:
+  nix-build:
+    runs-on: ubuntu-latest
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v6
+
+      - name: Install Nix
+        uses: DeterminateSystems/nix-installer-action@main
+
+      - name: Build flake package
+        run: |
+          nix build .#pangolin-newt -L

.github/workflows/nix-dependabot-update-hash.yml

@@ -0,0 +1,48 @@
+name: Update Nix Package Hash On Dependabot PRs
+
+on:
+  pull_request:
+    types: [opened, synchronize]
+    branches:
+      - main
+
+jobs:
+  nix-update:
+    if: github.actor == 'dependabot[bot]'
+    runs-on: ubuntu-latest
+
+    permissions:
+      contents: write
+
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@v6
+        with:
+          ref: ${{ github.head_ref }}
+          token: ${{ secrets.GITHUB_TOKEN }}
+
+      - name: Install Nix
+        uses: DeterminateSystems/nix-installer-action@main
+
+      - name: Run nix-update
+        run: |
+          nix run nixpkgs#nix-update -- --flake pangolin-newt --no-src --version skip
+
+      - name: Check for changes
+        id: changes
+        run: |
+          if git diff --quiet; then
+            echo "changed=false" >> "$GITHUB_OUTPUT"
+          else
+            echo "changed=true" >> "$GITHUB_OUTPUT"
+          fi
+
+      - name: Commit and push changes
+        if: steps.changes.outputs.changed == 'true'
+        run: |
+          git config user.name "dependabot[bot]"
+          git config user.email "dependabot[bot]@users.noreply.github.com"
+
+          git add .
+          git commit -m "chore(nix): fix hash for updated go dependencies"
+          git push

.github/workflows/stale-bot.yml

@@ -0,0 +1,37 @@
+name: Mark and Close Stale Issues
+
+on:
+  schedule:
+    - cron: '0 0 * * *'
+  workflow_dispatch: # Allow manual trigger
+
+permissions:
+  contents: write # only for delete-branch option
+  issues: write
+  pull-requests: write
+
+jobs:
+  stale:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/stale@eb5cf3af3ac0a1aa4c9c45633dd1ae542a27a899 # v10.3.0
+        with:
+          days-before-stale: 14
+          days-before-close: 14
+          stale-issue-message: 'This issue has been automatically marked as stale due to 14 days of inactivity. It will be closed in 14 days if no further activity occurs.'
+          close-issue-message: 'This issue has been automatically closed due to inactivity. If you believe this is still relevant, please open a new issue with up-to-date information.'
+          stale-issue-label: 'stale'
+          
+          exempt-issue-labels: 'needs investigating, networking, new feature, reverse proxy, bug, api, authentication, documentation, enhancement, help wanted, good first issue, question'
+          
+          exempt-all-issue-assignees: true
+          
+          only-labels: ''
+          exempt-pr-labels: ''
+          days-before-pr-stale: -1
+          days-before-pr-close: -1
+          
+          operations-per-run: 100
+          remove-stale-when-updated: true
+          delete-branch: false
+          enable-statistics: true

.github/workflows/test.yml

@@ -0,0 +1,53 @@
+name: Run Tests
+
+permissions:
+  contents: read
+
+on:
+  pull_request:
+    branches:
+      - main
+      - dev
+
+jobs:
+  test:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@8e8c483db84b4bee98b60c0593521ed34d9990e8 # v6.0.1
+
+      - name: Set up Go
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
+        with:
+          go-version-file: go.mod
+
+      - name: Run Go tests
+        run: make test
+
+  build:
+    runs-on: ubuntu-latest
+    strategy:
+      matrix:
+        target:
+          - local
+          - docker-build
+          - go-build-release-darwin-amd64
+          - go-build-release-darwin-arm64
+          - go-build-release-freebsd-amd64
+          - go-build-release-freebsd-arm64
+          - go-build-release-linux-amd64
+          - go-build-release-linux-arm32-v6
+          - go-build-release-linux-arm32-v7
+          - go-build-release-linux-riscv64
+          - go-build-release-windows-amd64
+    steps:
+      - name: Checkout repository
+        uses: actions/checkout@8e8c483db84b4bee98b60c0593521ed34d9990e8 # v6.0.1
+
+      - name: Set up Go
+        uses: actions/setup-go@4a3601121dd01d1626a1e23e37211e3254c1c06c # v6.4.0
+        with:
+          go-version-file: go.mod
+
+      - name: Build targets via `make`
+        run: make ${{ matrix.target }}

.gitignore

@@ -1,3 +1,10 @@
-newt
 .DS_Store
-bin/
+bin/
+nohup.out
+.idea
+*.iml
+certs/
+newt_arm64
+key
+/.direnv/
+/result*

.go-version

@@ -1 +1 @@
-1.23.2
+1.25

CONTRIBUTING.md

@@ -4,11 +4,7 @@ Contributions are welcome!
 
 Please see the contribution and local development guide on the docs page before getting started:
 
-https://docs.fossorial.io/development
-
-For ideas about what features to work on and our future plans, please see the roadmap:
-
-https://docs.fossorial.io/roadmap
+https://docs.pangolin.net/development/contributing
 
 ### Licensing Considerations
 
@@ -21,4 +17,4 @@ By creating this pull request, I grant the project maintainers an unlimited,
 perpetual license to use, modify, and redistribute these contributions under any terms they
 choose, including both the AGPLv3 and the Fossorial Commercial license terms. I
 represent that I have the right to grant this license for all contributed content.
-```
+```

Dockerfile

@@ -1,4 +1,8 @@
-FROM golang:1.23.1-alpine AS builder
+# FROM golang:1.25-alpine AS builder
+FROM public.ecr.aws/docker/library/golang:1.25-alpine AS builder
+
+# Install git and ca-certificates
+RUN apk --no-cache add ca-certificates git tzdata
 
 # Set the working directory inside the container
 WORKDIR /app
@@ -13,15 +17,23 @@ RUN go mod download
 COPY . .
 
 # Build the application
-RUN CGO_ENABLED=0 GOOS=linux go build -o /newt
+ARG VERSION=dev
+RUN CGO_ENABLED=0 GOOS=linux go build -ldflags="-s -w -X main.newtVersion=${VERSION}" -o /newt
 
-FROM alpine:3.19 AS runner
+FROM public.ecr.aws/docker/library/alpine:3.23 AS runner
 
-RUN apk --no-cache add ca-certificates
+RUN apk --no-cache add ca-certificates tzdata iputils
 
 COPY --from=builder /newt /usr/local/bin/
 COPY entrypoint.sh /
 
+# Marks this as an official Fossorial container image.
+# Auto-update is disabled in official images — update by pulling a new image tag.
+ENV NEWT_SYSTEM_SUBSTRATE="CONTAINER"
+
+# Admin/metrics endpoint (Prometheus scrape)
+EXPOSE 2112
+
 RUN chmod +x /entrypoint.sh
 ENTRYPOINT ["/entrypoint.sh"]
-CMD ["newt"]
+CMD ["newt"]

Makefile

@@ -1,37 +1,76 @@
+.PHONY: all local test docker-build docker-build-release
 
-all: build push
+all: local
+
+VERSION ?= dev
+LDFLAGS = -X main.newtVersion=$(VERSION)
+
+local:
+	CGO_ENABLED=0 go build -ldflags "$(LDFLAGS) -X main.newtPlatform=$(shell go env GOOS)_$(shell go env GOARCH)" -o ./bin/newt
+
+test:
+	go test ./...
+
+docker-build:
+	docker build -t fosrl/newt:latest .
 
 docker-build-release:
 	@if [ -z "$(tag)" ]; then \
-		echo "Error: tag is required. Usage: make build-all tag=<tag>"; \
+		echo "Error: tag is required. Usage: make docker-build-release tag=<tag>"; \
 		exit 1; \
 	fi
-	docker buildx build --platform linux/arm/v7,linux/arm64,linux/amd64 -t fosrl/newt:latest -f Dockerfile --push .
-	docker buildx build --platform linux/arm/v7,linux/arm64,linux/amd64 -t fosrl/newt:$(tag) -f Dockerfile --push .
+	docker buildx build . \
+		--platform linux/arm/v7,linux/arm64,linux/amd64 \
+		-t fosrl/newt:latest \
+		-t fosrl/newt:$(tag) \
+		-f Dockerfile \
+		--push
 
-build:
-	docker build -t fosrl/newt:latest .
+.PHONY: go-build-release \
+        go-build-release-linux-arm64 go-build-release-linux-arm32-v7 \
+        go-build-release-linux-arm32-v6 go-build-release-linux-amd64 \
+        go-build-release-linux-riscv64 go-build-release-darwin-arm64 \
+        go-build-release-darwin-amd64 go-build-release-windows-amd64 \
+        go-build-release-freebsd-amd64 go-build-release-freebsd-arm64
 
-push:
-	docker push fosrl/newt:latest
+go-build-release: \
+    go-build-release-linux-arm64 \
+    go-build-release-linux-arm32-v7 \
+    go-build-release-linux-arm32-v6 \
+    go-build-release-linux-amd64 \
+    go-build-release-linux-riscv64 \
+    go-build-release-darwin-arm64 \
+    go-build-release-darwin-amd64 \
+    go-build-release-windows-amd64 \
+    go-build-release-freebsd-amd64 \
+    go-build-release-freebsd-arm64
 
-test:
-	docker run fosrl/newt:latest
+go-build-release-linux-arm64:
+	CGO_ENABLED=0 GOOS=linux GOARCH=arm64 go build -ldflags "$(LDFLAGS) -X main.newtPlatform=linux_arm64" -o bin/newt_linux_arm64
 
-local: 
-	CGO_ENABLED=0 go build -o newt
+go-build-release-linux-arm32-v7:
+	CGO_ENABLED=0 GOOS=linux GOARCH=arm GOARM=7 go build -ldflags "$(LDFLAGS) -X main.newtPlatform=linux_arm32" -o bin/newt_linux_arm32
 
-go-build-release:
-	CGO_ENABLED=0 GOOS=linux GOARCH=arm64 go build -o bin/newt_linux_arm64
-	CGO_ENABLED=0 GOOS=linux GOARCH=arm GOARM=7 go build -o bin/newt_linux_arm32
-	CGO_ENABLED=0 GOOS=linux GOARCH=arm GOARM=6 go build -o bin/newt_linux_arm32v6
-	CGO_ENABLED=0 GOOS=linux GOARCH=amd64 go build -o bin/newt_linux_amd64
-	CGO_ENABLED=0 GOOS=linux GOARCH=riscv64 go build -o bin/newt_linux_riscv64
-	CGO_ENABLED=0 GOOS=darwin GOARCH=arm64 go build -o bin/newt_darwin_arm64
-	CGO_ENABLED=0 GOOS=darwin GOARCH=amd64 go build -o bin/newt_darwin_amd64
-	CGO_ENABLED=0 GOOS=windows GOARCH=amd64 go build -o bin/newt_windows_amd64.exe
-	CGO_ENABLED=0 GOOS=freebsd GOARCH=amd64 go build -o bin/newt_freebsd_amd64
-	CGO_ENABLED=0 GOOS=freebsd GOARCH=arm64 go build -o bin/newt_freebsd_arm64
+go-build-release-linux-arm32-v6:
+	CGO_ENABLED=0 GOOS=linux GOARCH=arm GOARM=6 go build -ldflags "$(LDFLAGS) -X main.newtPlatform=linux_arm32v6" -o bin/newt_linux_arm32v6
 
-clean:
-	rm newt
+go-build-release-linux-amd64:
+	CGO_ENABLED=0 GOOS=linux GOARCH=amd64 go build -ldflags "$(LDFLAGS) -X main.newtPlatform=linux_amd64" -o bin/newt_linux_amd64
+
+go-build-release-linux-riscv64:
+	CGO_ENABLED=0 GOOS=linux GOARCH=riscv64 go build -ldflags "$(LDFLAGS) -X main.newtPlatform=linux_riscv64" -o bin/newt_linux_riscv64
+
+go-build-release-darwin-arm64:
+	CGO_ENABLED=0 GOOS=darwin GOARCH=arm64 go build -ldflags "$(LDFLAGS) -X main.newtPlatform=darwin_arm64" -o bin/newt_darwin_arm64
+
+go-build-release-darwin-amd64:
+	CGO_ENABLED=0 GOOS=darwin GOARCH=amd64 go build -ldflags "$(LDFLAGS) -X main.newtPlatform=darwin_amd64" -o bin/newt_darwin_amd64
+
+go-build-release-windows-amd64:
+	CGO_ENABLED=0 GOOS=windows GOARCH=amd64 go build -ldflags "$(LDFLAGS) -X main.newtPlatform=windows_amd64" -o bin/newt_windows_amd64.exe
+
+go-build-release-freebsd-amd64:
+	CGO_ENABLED=0 GOOS=freebsd GOARCH=amd64 go build -ldflags "$(LDFLAGS) -X main.newtPlatform=freebsd_amd64" -o bin/newt_freebsd_amd64
+
+go-build-release-freebsd-arm64:
+	CGO_ENABLED=0 GOOS=freebsd GOARCH=arm64 go build -ldflags "$(LDFLAGS) -X main.newtPlatform=freebsd_arm64" -o bin/newt_freebsd_arm64

README.md

@@ -1,4 +1,7 @@
 # Newt
+[![PkgGoDev](https://pkg.go.dev/badge/github.com/fosrl/newt)](https://pkg.go.dev/github.com/fosrl/newt)
+[![GitHub License](https://img.shields.io/github/license/fosrl/newt)](https://github.com/fosrl/newt/blob/main/LICENSE)
+[![Go Report Card](https://goreportcard.com/badge/github.com/fosrl/newt)](https://goreportcard.com/report/github.com/fosrl/newt)
 
 Newt is a fully user space [WireGuard](https://www.wireguard.com/) tunnel client and TCP/UDP proxy, designed to securely expose private resources controlled by Pangolin. By using Newt, you don't need to manage complex WireGuard tunnels and NATing.
 
@@ -6,14 +9,7 @@ Newt is a fully user space [WireGuard](https://www.wireguard.com/) tunnel client
 
 Newt is used with Pangolin and Gerbil as part of the larger system. See documentation below:
 
--   [Installation Instructions](https://docs.fossorial.io)
--   [Full Documentation](https://docs.fossorial.io)
-
-## Preview
-
-<img src="public/screenshots/preview.png" alt="Preview"/>
-
-_Sample output of a Newt container connected to Pangolin and hosting various resource target proxies._
+-   [Full Documentation](https://docs.pangolin.net/manage/sites/understanding-sites)
 
 ## Key Functions
 
@@ -23,108 +19,32 @@ Using the Newt ID and a secret, the client will make HTTP requests to Pangolin t
 
 ### Receives WireGuard Control Messages
 
-When Newt receives WireGuard control messages, it will use the information encoded (endpoint, public key) to bring up a WireGuard tunnel using [netstack](https://github.com/WireGuard/wireguard-go/blob/master/tun/netstack/examples/http_server.go) fully in user space. It will ping over the tunnel to ensure the peer on the Gerbil side is brought up. 
+When Newt receives WireGuard control messages, it will use the information encoded (endpoint, public key) to bring up a WireGuard tunnel using [netstack](https://github.com/WireGuard/wireguard-go/blob/master/tun/netstack/examples/http_server.go) fully in user space. It will ping over the tunnel to ensure the peer on the Gerbil side is brought up.
 
 ### Receives Proxy Control Messages
 
 When Newt receives WireGuard control messages, it will use the information encoded to create a local low level TCP and UDP proxies attached to the virtual tunnel in order to relay traffic to programmed targets.
 
-## CLI Args
-
-- `endpoint`: The endpoint where both Gerbil and Pangolin reside in order to connect to the websocket.
-- `id`: Newt ID generated by Pangolin to identify the client.
-- `secret`: A unique secret (not shared and kept private) used to authenticate the client ID with the websocket in order to receive commands. 
-- `dns`: DNS server to use to resolve the endpoint
-- `log-level` (optional): The log level to use. Default: INFO
-- `updown` (optional): A script to be called when targets are added or removed.
- 
-Example:
-
-```bash
-./newt \
---id 31frd0uzbjvp721 \
---secret h51mmlknrvrwv8s4r1i210azhumt6isgbpyavxodibx1k2d6 \
---endpoint https://example.com
-```
-
-You can also run it with Docker compose. For example, a service in your `docker-compose.yml` might look like this using environment vars (recommended):
-
-```yaml
-services:
-  newt:
-    image: fosrl/newt
-    container_name: newt
-    restart: unless-stopped
-    environment:
-      - PANGOLIN_ENDPOINT=https://example.com
-      - NEWT_ID=2ix2t8xk22ubpfy 
-      - NEWT_SECRET=nnisrfsdfc7prqsp9ewo1dvtvci50j5uiqotez00dgap0ii2 
-```
-
-You can also pass the CLI args to the container:
-
-```yaml
-services:
-  newt:
-    image: fosrl/newt
-    container_name: newt
-    restart: unless-stopped
-    command:
-        - --id 31frd0uzbjvp721
-        - --secret h51mmlknrvrwv8s4r1i210azhumt6isgbpyavxodibx1k2d6
-        - --endpoint https://example.com
-```
-
-Finally a basic systemd service:
-
-```
-[Unit]
-Description=Newt VPN Client
-After=network.target
-
-[Service]
-ExecStart=/usr/local/bin/newt --id 31frd0uzbjvp721 --secret h51mmlknrvrwv8s4r1i210azhumt6isgbpyavxodibx1k2d6 --endpoint https://example.com
-Restart=always
-User=root
-
-[Install]
-WantedBy=multi-user.target
-```
-
-Make sure to `mv ./newt /usr/local/bin/newt`!
-
-### Updown
-
-You can pass in a updown script for Newt to call when it is adding or removing a target:
-
-`--updown "python3 test.py"`
-
-It will get called with args when a target is added: 
-`python3 test.py add tcp localhost:8556`
-`python3 test.py remove tcp localhost:8556`
-
-Returning a string from the script in the format of a target (`ip:dst` so `10.0.0.1:8080`) it will override the target and use this value instead to proxy.
-
-You can look at updown.py as a reference script to get started!
-
 ## Build
 
-### Container 
+### Binary
 
-Ensure Docker is installed.
+Make sure to have Go 1.25 installed.
 
 ```bash
 make
 ```
 
-### Binary
-
-Make sure to have Go 1.23.1 installed.
+### Nix Flake
 
 ```bash
-make local
+nix build
 ```
 
+Binary will be at `./result/bin/newt`
+
+Development shell available with `nix develop`
+
 ## Licensing
 
 Newt is dual licensed under the AGPLv3 and the Fossorial Commercial license. For inquiries about commercial licensing, please contact us.

SECURITY.md

@@ -3,7 +3,7 @@
 If you discover a security vulnerability, please follow the steps below to responsibly disclose it to us:
 
 1. **Do not create a public GitHub issue or discussion post.** This could put the security of other users at risk.
-2. Send a detailed report to [security@fossorial.io](mailto:security@fossorial.io) or send a **private** message to a maintainer on [Discord](https://discord.gg/HCJR8Xhme4). Include:
+2. Send a detailed report to [security@pangolin.net](mailto:security@pangolin.net) or send a **private** message to a maintainer on [Discord](https://discord.gg/HCJR8Xhme4). Include:
 
 - Description and location of the vulnerability.
 - Potential impact of the vulnerability.

authdaemon.go

@@ -0,0 +1,150 @@
+package main
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"os"
+	"runtime"
+
+	"github.com/fosrl/newt/authdaemon"
+	"github.com/fosrl/newt/logger"
+)
+
+const (
+	defaultPrincipalsPath = "/var/run/auth-daemon/principals"
+	defaultCACertPath     = "/etc/ssh/ca.pem"
+)
+
+var (
+	errPresharedKeyRequired = errors.New("auth-daemon-key is required when --auth-daemon is enabled")
+	errRootRequired         = errors.New("auth-daemon must be run as root (use sudo)")
+	authDaemonServer        *authdaemon.Server // Global auth daemon server instance
+)
+
+// startAuthDaemon initializes and starts the auth daemon in the background.
+// It validates requirements (Linux, root, preshared key) and starts the server
+// in a goroutine so it runs alongside normal newt operation.
+func startAuthDaemon(ctx context.Context) error {
+	// Validation
+	if runtime.GOOS != "linux" {
+		return fmt.Errorf("auth-daemon is only supported on Linux, not %s", runtime.GOOS)
+	}
+	if os.Geteuid() != 0 {
+		return errRootRequired
+	}
+
+	// Use defaults if not set
+	principalsFile := authDaemonPrincipalsFile
+	if principalsFile == "" {
+		principalsFile = defaultPrincipalsPath
+	}
+	caCertPath := authDaemonCACertPath
+	if caCertPath == "" {
+		caCertPath = defaultCACertPath
+	}
+
+	// Create auth daemon server
+	cfg := authdaemon.Config{
+		DisableHTTPS:           true,                   // We run without HTTP server in newt
+		PresharedKey:           "this-key-is-not-used", // Not used in embedded mode, but set to non-empty to satisfy validation
+		PrincipalsFilePath:     principalsFile,
+		CACertPath:             caCertPath,
+		Force:                  true,
+		GenerateRandomPassword: authDaemonGenerateRandomPassword,
+	}
+
+	srv, err := authdaemon.NewServer(cfg)
+	if err != nil {
+		return fmt.Errorf("create auth daemon server: %w", err)
+	}
+
+	authDaemonServer = srv
+
+	// Start the auth daemon in a goroutine so it runs alongside newt
+	go func() {
+		logger.Debug("Auth daemon starting (native mode, no HTTP server)")
+		if err := srv.Run(ctx); err != nil {
+			logger.Error("Auth daemon error: %v", err)
+		}
+		logger.Info("Auth daemon stopped")
+	}()
+
+	return nil
+}
+
+// runPrincipalsCmd executes the principals subcommand logic
+func runPrincipalsCmd(args []string) {
+	opts := struct {
+		PrincipalsFile string
+		Username       string
+	}{
+		PrincipalsFile: defaultPrincipalsPath,
+	}
+
+	// Parse flags manually
+	for i := 0; i < len(args); i++ {
+		switch args[i] {
+		case "--principals-file":
+			if i+1 >= len(args) {
+				fmt.Fprintf(os.Stderr, "Error: --principals-file requires a value\n")
+				os.Exit(1)
+			}
+			opts.PrincipalsFile = args[i+1]
+			i++
+		case "--username":
+			if i+1 >= len(args) {
+				fmt.Fprintf(os.Stderr, "Error: --username requires a value\n")
+				os.Exit(1)
+			}
+			opts.Username = args[i+1]
+			i++
+		case "--help", "-h":
+			printPrincipalsHelp()
+			os.Exit(0)
+		default:
+			fmt.Fprintf(os.Stderr, "Error: unknown flag: %s\n", args[i])
+			printPrincipalsHelp()
+			os.Exit(1)
+		}
+	}
+
+	// Validation
+	if opts.Username == "" {
+		fmt.Fprintf(os.Stderr, "Error: username is required\n")
+		printPrincipalsHelp()
+		os.Exit(1)
+	}
+
+	// Get principals
+	list, err := authdaemon.GetPrincipals(opts.PrincipalsFile, opts.Username)
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "Error: %v\n", err)
+		os.Exit(1)
+	}
+	if len(list) == 0 {
+		fmt.Println("")
+		return
+	}
+	for _, principal := range list {
+		fmt.Println(principal)
+	}
+}
+
+func printPrincipalsHelp() {
+	fmt.Fprintf(os.Stderr, `Usage: newt principals [flags]
+
+Output principals for a username (for AuthorizedPrincipalsCommand in sshd_config).
+Read the principals file and print principals that match the given username, one per line.
+Configure in sshd_config with AuthorizedPrincipalsCommand and %%u for the username.
+
+Flags:
+  --principals-file string   Path to the principals file (default "%s")
+  --username string          Username to look up (required)
+  --help, -h                 Show this help message
+
+Example:
+  newt principals --username alice
+
+`, defaultPrincipalsPath)
+}

authdaemon/connection.go

@@ -0,0 +1,27 @@
+package authdaemon
+
+import (
+	"github.com/fosrl/newt/logger"
+)
+
+// ProcessConnection runs the same logic as POST /connection: CA cert, user create/reconcile, principals.
+// Use this when DisableHTTPS is true (e.g. embedded in Newt) instead of calling the API.
+func (s *Server) ProcessConnection(req ConnectionRequest) {
+	logger.Info("connection: niceId=%q username=%q metadata.sudoMode=%q metadata.sudoCommands=%v metadata.homedir=%v metadata.groups=%v",
+		req.NiceId, req.Username, req.Metadata.SudoMode, req.Metadata.SudoCommands, req.Metadata.Homedir, req.Metadata.Groups)
+
+	cfg := &s.cfg
+	if cfg.CACertPath != "" {
+		if err := writeCACertIfNotExists(cfg.CACertPath, req.CaCert, cfg.Force); err != nil {
+			logger.Warn("auth-daemon: write CA cert: %v", err)
+		}
+	}
+	if err := ensureUser(req.Username, req.Metadata, s.cfg.GenerateRandomPassword); err != nil {
+		logger.Warn("auth-daemon: ensure user: %v", err)
+	}
+	if cfg.PrincipalsFilePath != "" && req.NiceId != "" {
+		if err := writePrincipals(cfg.PrincipalsFilePath, req.Username, req.NiceId); err != nil {
+			logger.Warn("auth-daemon: write principals: %v", err)
+		}
+	}
+}

authdaemon/host_linux.go

@@ -0,0 +1,396 @@
+//go:build linux
+
+package authdaemon
+
+import (
+	"bufio"
+	"crypto/rand"
+	"encoding/hex"
+	"encoding/json"
+	"fmt"
+	"os"
+	"os/exec"
+	"os/user"
+	"path/filepath"
+	"strconv"
+	"strings"
+
+	"github.com/fosrl/newt/logger"
+)
+
+// writeCACertIfNotExists writes contents to path. If the file already exists: when force is false, skip; when force is true, overwrite only if content differs.
+func writeCACertIfNotExists(path, contents string, force bool) error {
+	contents = strings.TrimSpace(contents)
+	if contents != "" && !strings.HasSuffix(contents, "\n") {
+		contents += "\n"
+	}
+	existing, err := os.ReadFile(path)
+	if err == nil {
+		existingStr := strings.TrimSpace(string(existing))
+		if existingStr != "" && !strings.HasSuffix(existingStr, "\n") {
+			existingStr += "\n"
+		}
+		if existingStr == contents {
+			logger.Debug("auth-daemon: CA cert unchanged at %s, skipping write", path)
+			return nil
+		}
+		if !force {
+			logger.Debug("auth-daemon: CA cert already exists at %s, skipping write (Force disabled)", path)
+			return nil
+		}
+	} else if !os.IsNotExist(err) {
+		return fmt.Errorf("read %s: %w", path, err)
+	}
+	dir := filepath.Dir(path)
+	if err := os.MkdirAll(dir, 0755); err != nil {
+		return fmt.Errorf("mkdir %s: %w", dir, err)
+	}
+	if err := os.WriteFile(path, []byte(contents), 0644); err != nil {
+		return fmt.Errorf("write CA cert: %w", err)
+	}
+	logger.Info("auth-daemon: wrote CA cert to %s", path)
+	return nil
+}
+
+// writePrincipals updates the principals file at path: JSON object keyed by username, value is array of principals. Adds username and niceId to that user's list (deduped).
+func writePrincipals(path, username, niceId string) error {
+	if path == "" {
+		return nil
+	}
+	username = strings.TrimSpace(username)
+	niceId = strings.TrimSpace(niceId)
+	if username == "" {
+		return nil
+	}
+	data := make(map[string][]string)
+	if raw, err := os.ReadFile(path); err == nil {
+		_ = json.Unmarshal(raw, &data)
+	}
+	list := data[username]
+	seen := make(map[string]struct{}, len(list)+2)
+	for _, p := range list {
+		seen[p] = struct{}{}
+	}
+	for _, p := range []string{username, niceId} {
+		if p == "" {
+			continue
+		}
+		if _, ok := seen[p]; !ok {
+			seen[p] = struct{}{}
+			list = append(list, p)
+		}
+	}
+	data[username] = list
+	body, err := json.Marshal(data)
+	if err != nil {
+		return fmt.Errorf("marshal principals: %w", err)
+	}
+	dir := filepath.Dir(path)
+	if err := os.MkdirAll(dir, 0755); err != nil {
+		return fmt.Errorf("mkdir %s: %w", dir, err)
+	}
+	if err := os.WriteFile(path, body, 0644); err != nil {
+		return fmt.Errorf("write principals: %w", err)
+	}
+	logger.Debug("auth-daemon: wrote principals to %s", path)
+	return nil
+}
+
+// sudoGroup returns the name of the sudo group (wheel or sudo) that exists on the system. Prefers wheel.
+func sudoGroup() string {
+	f, err := os.Open("/etc/group")
+	if err != nil {
+		return "sudo"
+	}
+	defer f.Close()
+	sc := bufio.NewScanner(f)
+	hasWheel := false
+	hasSudo := false
+	for sc.Scan() {
+		line := sc.Text()
+		if strings.HasPrefix(line, "wheel:") {
+			hasWheel = true
+		}
+		if strings.HasPrefix(line, "sudo:") {
+			hasSudo = true
+		}
+	}
+	if hasWheel {
+		return "wheel"
+	}
+	if hasSudo {
+		return "sudo"
+	}
+	return "sudo"
+}
+
+// setRandomPassword generates a random password and sets it for username via chpasswd.
+// Used when GenerateRandomPassword is true so SSH with PermitEmptyPasswords no can accept the user.
+func setRandomPassword(username string) error {
+	b := make([]byte, 16)
+	if _, err := rand.Read(b); err != nil {
+		return fmt.Errorf("generate password: %w", err)
+	}
+	password := hex.EncodeToString(b)
+	cmd := exec.Command("chpasswd")
+	cmd.Stdin = strings.NewReader(username + ":" + password)
+	if out, err := cmd.CombinedOutput(); err != nil {
+		return fmt.Errorf("chpasswd: %w (output: %s)", err, string(out))
+	}
+	return nil
+}
+
+const skelDir = "/etc/skel"
+
+// copySkelInto copies files from srcDir (e.g. /etc/skel) into dstDir (e.g. user's home).
+// Only creates files that don't already exist. All created paths are chowned to uid:gid.
+func copySkelInto(srcDir, dstDir string, uid, gid int) {
+	entries, err := os.ReadDir(srcDir)
+	if err != nil {
+		if !os.IsNotExist(err) {
+			logger.Warn("auth-daemon: read %s: %v", srcDir, err)
+		}
+		return
+	}
+	for _, e := range entries {
+		name := e.Name()
+		src := filepath.Join(srcDir, name)
+		dst := filepath.Join(dstDir, name)
+		if e.IsDir() {
+			if st, err := os.Stat(dst); err == nil && st.IsDir() {
+				copySkelInto(src, dst, uid, gid)
+				continue
+			}
+			if err := os.MkdirAll(dst, 0755); err != nil {
+				logger.Warn("auth-daemon: mkdir %s: %v", dst, err)
+				continue
+			}
+			if err := os.Chown(dst, uid, gid); err != nil {
+				logger.Warn("auth-daemon: chown %s: %v", dst, err)
+			}
+			copySkelInto(src, dst, uid, gid)
+			continue
+		}
+		if _, err := os.Stat(dst); err == nil {
+			continue
+		}
+		data, err := os.ReadFile(src)
+		if err != nil {
+			logger.Warn("auth-daemon: read %s: %v", src, err)
+			continue
+		}
+		if err := os.WriteFile(dst, data, 0644); err != nil {
+			logger.Warn("auth-daemon: write %s: %v", dst, err)
+			continue
+		}
+		if err := os.Chown(dst, uid, gid); err != nil {
+			logger.Warn("auth-daemon: chown %s: %v", dst, err)
+		}
+	}
+}
+
+// ensureUser creates the system user if missing, or reconciles sudo and homedir to match meta.
+func ensureUser(username string, meta ConnectionMetadata, generateRandomPassword bool) error {
+	if username == "" {
+		return nil
+	}
+	u, err := user.Lookup(username)
+	if err != nil {
+		if _, ok := err.(user.UnknownUserError); !ok {
+			return fmt.Errorf("lookup user %s: %w", username, err)
+		}
+		return createUser(username, meta, generateRandomPassword)
+	}
+	return reconcileUser(u, meta)
+}
+
+// desiredGroups returns the exact list of supplementary groups the user should have:
+// meta.Groups plus the sudo group when meta.SudoMode is "full" (deduped).
+func desiredGroups(meta ConnectionMetadata) []string {
+	seen := make(map[string]struct{})
+	var out []string
+	for _, g := range meta.Groups {
+		g = strings.TrimSpace(g)
+		if g == "" {
+			continue
+		}
+		if _, ok := seen[g]; ok {
+			continue
+		}
+		seen[g] = struct{}{}
+		out = append(out, g)
+	}
+	if meta.SudoMode == "full" {
+		sg := sudoGroup()
+		if _, ok := seen[sg]; !ok {
+			out = append(out, sg)
+		}
+	}
+	return out
+}
+
+// setUserGroups sets the user's supplementary groups to exactly groups (local mirrors metadata).
+// When groups is empty, clears all supplementary groups (usermod -G "").
+func setUserGroups(username string, groups []string) {
+	list := strings.Join(groups, ",")
+	cmd := exec.Command("usermod", "-G", list, username)
+	if out, err := cmd.CombinedOutput(); err != nil {
+		logger.Warn("auth-daemon: usermod -G %s: %v (output: %s)", list, err, string(out))
+	} else {
+		logger.Info("auth-daemon: set %s supplementary groups to %s", username, list)
+	}
+}
+
+func createUser(username string, meta ConnectionMetadata, generateRandomPassword bool) error {
+	args := []string{"-s", "/bin/bash"}
+	if meta.Homedir {
+		args = append(args, "-m")
+	} else {
+		args = append(args, "-M")
+	}
+	args = append(args, username)
+	cmd := exec.Command("useradd", args...)
+	if out, err := cmd.CombinedOutput(); err != nil {
+		return fmt.Errorf("useradd %s: %w (output: %s)", username, err, string(out))
+	}
+	logger.Info("auth-daemon: created user %s (homedir=%v)", username, meta.Homedir)
+	if generateRandomPassword {
+		if err := setRandomPassword(username); err != nil {
+			logger.Warn("auth-daemon: set random password for %s: %v", username, err)
+		} else {
+			logger.Info("auth-daemon: set random password for %s (PermitEmptyPasswords no)", username)
+		}
+	}
+	if meta.Homedir {
+		if u, err := user.Lookup(username); err == nil && u.HomeDir != "" {
+			uid, gid := mustAtoi(u.Uid), mustAtoi(u.Gid)
+			copySkelInto(skelDir, u.HomeDir, uid, gid)
+		}
+	}
+	setUserGroups(username, desiredGroups(meta))
+	switch meta.SudoMode {
+	case "full":
+		if err := configurePasswordlessSudo(username); err != nil {
+			logger.Warn("auth-daemon: configure passwordless sudo for %s: %v", username, err)
+		}
+	case "commands":
+		if len(meta.SudoCommands) > 0 {
+			if err := configureSudoCommands(username, meta.SudoCommands); err != nil {
+				logger.Warn("auth-daemon: configure sudo commands for %s: %v", username, err)
+			}
+		}
+	default:
+		removeSudoers(username)
+	}
+	return nil
+}
+
+const sudoersFilePrefix = "90-pangolin-"
+
+func sudoersPath(username string) string {
+	return filepath.Join("/etc/sudoers.d", sudoersFilePrefix+username)
+}
+
+// writeSudoersFile writes content to the user's sudoers.d file and validates with visudo.
+func writeSudoersFile(username, content string) error {
+	sudoersFile := sudoersPath(username)
+	tmpFile := sudoersFile + ".tmp"
+	if err := os.WriteFile(tmpFile, []byte(content), 0440); err != nil {
+		return fmt.Errorf("write temp sudoers file: %w", err)
+	}
+	cmd := exec.Command("visudo", "-c", "-f", tmpFile)
+	if out, err := cmd.CombinedOutput(); err != nil {
+		os.Remove(tmpFile)
+		return fmt.Errorf("visudo validation failed: %w (output: %s)", err, string(out))
+	}
+	if err := os.Rename(tmpFile, sudoersFile); err != nil {
+		os.Remove(tmpFile)
+		return fmt.Errorf("move sudoers file: %w", err)
+	}
+	return nil
+}
+
+// configurePasswordlessSudo creates a sudoers.d file to allow passwordless sudo for the user.
+func configurePasswordlessSudo(username string) error {
+	content := fmt.Sprintf("# Created by Pangolin auth-daemon\n%s ALL=(ALL) NOPASSWD:ALL\n", username)
+	if err := writeSudoersFile(username, content); err != nil {
+		return err
+	}
+	logger.Info("auth-daemon: configured passwordless sudo for %s", username)
+	return nil
+}
+
+// configureSudoCommands creates a sudoers.d file allowing only the listed commands (NOPASSWD).
+// Each command should be a full path (e.g. /usr/bin/systemctl).
+func configureSudoCommands(username string, commands []string) error {
+	var b strings.Builder
+	b.WriteString("# Created by Pangolin auth-daemon (restricted commands)\n")
+	n := 0
+	for _, c := range commands {
+		c = strings.TrimSpace(c)
+		if c == "" {
+			continue
+		}
+		fmt.Fprintf(&b, "%s ALL=(ALL) NOPASSWD: %s\n", username, c)
+		n++
+	}
+	if n == 0 {
+		return fmt.Errorf("no valid sudo commands")
+	}
+	if err := writeSudoersFile(username, b.String()); err != nil {
+		return err
+	}
+	logger.Info("auth-daemon: configured restricted sudo for %s (%d commands)", username, len(commands))
+	return nil
+}
+
+// removeSudoers removes the sudoers.d file for the user.
+func removeSudoers(username string) {
+	sudoersFile := sudoersPath(username)
+	if err := os.Remove(sudoersFile); err != nil && !os.IsNotExist(err) {
+		logger.Warn("auth-daemon: remove sudoers for %s: %v", username, err)
+	} else if err == nil {
+		logger.Info("auth-daemon: removed sudoers for %s", username)
+	}
+}
+
+func mustAtoi(s string) int {
+	n, _ := strconv.Atoi(s)
+	return n
+}
+
+func reconcileUser(u *user.User, meta ConnectionMetadata) error {
+	setUserGroups(u.Username, desiredGroups(meta))
+	switch meta.SudoMode {
+	case "full":
+		if err := configurePasswordlessSudo(u.Username); err != nil {
+			logger.Warn("auth-daemon: configure passwordless sudo for %s: %v", u.Username, err)
+		}
+	case "commands":
+		if len(meta.SudoCommands) > 0 {
+			if err := configureSudoCommands(u.Username, meta.SudoCommands); err != nil {
+				logger.Warn("auth-daemon: configure sudo commands for %s: %v", u.Username, err)
+			}
+		} else {
+			removeSudoers(u.Username)
+		}
+	default:
+		removeSudoers(u.Username)
+	}
+	if meta.Homedir && u.HomeDir != "" {
+		uid, gid := mustAtoi(u.Uid), mustAtoi(u.Gid)
+		if st, err := os.Stat(u.HomeDir); err != nil || !st.IsDir() {
+			if err := os.MkdirAll(u.HomeDir, 0755); err != nil {
+				logger.Warn("auth-daemon: mkdir %s: %v", u.HomeDir, err)
+			} else {
+				_ = os.Chown(u.HomeDir, uid, gid)
+				copySkelInto(skelDir, u.HomeDir, uid, gid)
+				logger.Info("auth-daemon: created home %s for %s", u.HomeDir, u.Username)
+			}
+		} else {
+			// Ensure .bashrc etc. exist (e.g. home existed but was empty or skel was minimal)
+			copySkelInto(skelDir, u.HomeDir, uid, gid)
+		}
+	}
+	return nil
+}

authdaemon/host_stub.go

@@ -0,0 +1,22 @@
+//go:build !linux
+
+package authdaemon
+
+import "fmt"
+
+var errLinuxOnly = fmt.Errorf("auth-daemon PAM agent is only supported on Linux")
+
+// writeCACertIfNotExists returns an error on non-Linux.
+func writeCACertIfNotExists(path, contents string, force bool) error {
+	return errLinuxOnly
+}
+
+// ensureUser returns an error on non-Linux.
+func ensureUser(username string, meta ConnectionMetadata, generateRandomPassword bool) error {
+	return errLinuxOnly
+}
+
+// writePrincipals returns an error on non-Linux.
+func writePrincipals(path, username, niceId string) error {
+	return errLinuxOnly
+}

authdaemon/principals.go

@@ -0,0 +1,28 @@
+package authdaemon
+
+import (
+	"encoding/json"
+	"fmt"
+	"os"
+)
+
+// GetPrincipals reads the principals data file at path, looks up the given user, and returns that user's principals as a string slice.
+// The file format is JSON: object with username keys and array-of-principals values, e.g. {"alice":["alice","usr-123"],"bob":["bob","usr-456"]}.
+// If the user is not found or the file is missing, returns nil and nil.
+func GetPrincipals(path, user string) ([]string, error) {
+	if path == "" {
+		return nil, fmt.Errorf("principals file path is required")
+	}
+	data, err := os.ReadFile(path)
+	if err != nil {
+		if os.IsNotExist(err) {
+			return nil, nil
+		}
+		return nil, fmt.Errorf("read principals file: %w", err)
+	}
+	var m map[string][]string
+	if err := json.Unmarshal(data, &m); err != nil {
+		return nil, fmt.Errorf("parse principals file: %w", err)
+	}
+	return m[user], nil
+}

authdaemon/routes.go

@@ -0,0 +1,58 @@
+package authdaemon
+
+import (
+	"encoding/json"
+	"net/http"
+)
+
+// registerRoutes registers all API routes. Add new endpoints here.
+func (s *Server) registerRoutes() {
+	s.mux.HandleFunc("/health", s.handleHealth)
+	s.mux.HandleFunc("/connection", s.handleConnection)
+}
+
+// ConnectionMetadata is the metadata object in POST /connection.
+type ConnectionMetadata struct {
+	SudoMode     string   `json:"sudoMode"`     // "none" | "full" | "commands"
+	SudoCommands []string `json:"sudoCommands"` // used when sudoMode is "commands"
+	Homedir      bool     `json:"homedir"`
+	Groups       []string `json:"groups"` // system groups to add the user to
+}
+
+// ConnectionRequest is the JSON body for POST /connection.
+type ConnectionRequest struct {
+	CaCert   string             `json:"caCert"`
+	NiceId   string             `json:"niceId"`
+	Username string             `json:"username"`
+	Metadata ConnectionMetadata `json:"metadata"`
+}
+
+// healthResponse is the JSON body for GET /health.
+type healthResponse struct {
+	Status string `json:"status"`
+}
+
+// handleHealth responds with 200 and {"status":"ok"}.
+func (s *Server) handleHealth(w http.ResponseWriter, r *http.Request) {
+	if r.Method != http.MethodGet {
+		http.Error(w, "Method Not Allowed", http.StatusMethodNotAllowed)
+		return
+	}
+	w.Header().Set("Content-Type", "application/json")
+	_ = json.NewEncoder(w).Encode(healthResponse{Status: "ok"})
+}
+
+// handleConnection accepts POST with connection payload and delegates to ProcessConnection.
+func (s *Server) handleConnection(w http.ResponseWriter, r *http.Request) {
+	if r.Method != http.MethodPost {
+		http.Error(w, "Method Not Allowed", http.StatusMethodNotAllowed)
+		return
+	}
+	var req ConnectionRequest
+	if err := json.NewDecoder(r.Body).Decode(&req); err != nil {
+		http.Error(w, "Bad Request", http.StatusBadRequest)
+		return
+	}
+	s.ProcessConnection(req)
+	w.WriteHeader(http.StatusOK)
+}

authdaemon/server.go

@@ -0,0 +1,180 @@
+package authdaemon
+
+import (
+	"context"
+	"crypto/ecdsa"
+	"crypto/elliptic"
+	"crypto/rand"
+	"crypto/subtle"
+	"crypto/tls"
+	"crypto/x509"
+	"crypto/x509/pkix"
+	"encoding/pem"
+	"fmt"
+	"math/big"
+	"net/http"
+	"os"
+	"runtime"
+	"strings"
+	"time"
+
+	"github.com/fosrl/newt/logger"
+)
+
+type Config struct {
+	// DisableHTTPS: when true, Run() does not start the HTTPS server (for embedded use inside Newt). Call ProcessConnection directly for connection events.
+	DisableHTTPS           bool
+	Port                   int    // Required when DisableHTTPS is false. Listen port for the HTTPS server. No default.
+	PresharedKey           string // Required when DisableHTTPS is false. HTTP auth (Authorization: Bearer <key> or X-Preshared-Key: <key>). No default.
+	CACertPath             string // Required. Where to write the CA cert (e.g. /etc/ssh/ca.pem). No default.
+	Force                  bool   // If true, overwrite existing CA cert (and other items) when content differs. Default false.
+	PrincipalsFilePath     string // Required. Path to the principals data file (JSON: username -> array of principals). No default.
+	GenerateRandomPassword bool   // If true, set a random password on users when they are provisioned (for SSH PermitEmptyPasswords no).
+}
+
+type Server struct {
+	cfg          Config
+	addr         string
+	presharedKey string
+	mux          *http.ServeMux
+	tlsCert      tls.Certificate
+}
+
+// generateTLSCert creates a self-signed certificate and key in memory (no disk).
+func generateTLSCert() (tls.Certificate, error) {
+	key, err := ecdsa.GenerateKey(elliptic.P256(), rand.Reader)
+	if err != nil {
+		return tls.Certificate{}, fmt.Errorf("generate key: %w", err)
+	}
+	serial, err := rand.Int(rand.Reader, new(big.Int).Lsh(big.NewInt(1), 128))
+	if err != nil {
+		return tls.Certificate{}, fmt.Errorf("serial: %w", err)
+	}
+	tmpl := &x509.Certificate{
+		SerialNumber: serial,
+		Subject: pkix.Name{
+			CommonName: "localhost",
+		},
+		NotBefore:             time.Now(),
+		NotAfter:              time.Now().Add(365 * 24 * time.Hour),
+		KeyUsage:              x509.KeyUsageDigitalSignature,
+		ExtKeyUsage:           []x509.ExtKeyUsage{x509.ExtKeyUsageServerAuth},
+		BasicConstraintsValid: true,
+		DNSNames:              []string{"localhost", "127.0.0.1"},
+	}
+	certDER, err := x509.CreateCertificate(rand.Reader, tmpl, tmpl, key.Public(), key)
+	if err != nil {
+		return tls.Certificate{}, fmt.Errorf("create certificate: %w", err)
+	}
+	keyDER, err := x509.MarshalECPrivateKey(key)
+	if err != nil {
+		return tls.Certificate{}, fmt.Errorf("marshal key: %w", err)
+	}
+	certPEM := pem.EncodeToMemory(&pem.Block{Type: "CERTIFICATE", Bytes: certDER})
+	keyPEM := pem.EncodeToMemory(&pem.Block{Type: "EC PRIVATE KEY", Bytes: keyDER})
+	cert, err := tls.X509KeyPair(certPEM, keyPEM)
+	if err != nil {
+		return tls.Certificate{}, fmt.Errorf("x509 key pair: %w", err)
+	}
+	return cert, nil
+}
+
+// authMiddleware wraps next and requires a valid preshared key on every request.
+// Accepts Authorization: Bearer <key> or X-Preshared-Key: <key>.
+func (s *Server) authMiddleware(next http.Handler) http.Handler {
+	return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		key := ""
+		if v := r.Header.Get("Authorization"); strings.HasPrefix(v, "Bearer ") {
+			key = strings.TrimSpace(strings.TrimPrefix(v, "Bearer "))
+		}
+		if key == "" {
+			key = strings.TrimSpace(r.Header.Get("X-Preshared-Key"))
+		}
+		if key == "" || subtle.ConstantTimeCompare([]byte(key), []byte(s.presharedKey)) != 1 {
+			http.Error(w, "Unauthorized", http.StatusUnauthorized)
+			return
+		}
+		next.ServeHTTP(w, r)
+	})
+}
+
+// NewServer builds a new auth-daemon server from cfg. Port, PresharedKey, CACertPath, and PrincipalsFilePath are required (no defaults).
+func NewServer(cfg Config) (*Server, error) {
+	if runtime.GOOS != "linux" {
+		return nil, fmt.Errorf("auth-daemon is only supported on Linux, not %s", runtime.GOOS)
+	}
+	if !cfg.DisableHTTPS {
+		if cfg.Port <= 0 {
+			return nil, fmt.Errorf("port is required and must be positive")
+		}
+		if cfg.PresharedKey == "" {
+			return nil, fmt.Errorf("preshared key is required")
+		}
+	}
+	if cfg.CACertPath == "" {
+		return nil, fmt.Errorf("CACertPath is required")
+	}
+	if cfg.PrincipalsFilePath == "" {
+		return nil, fmt.Errorf("PrincipalsFilePath is required")
+	}
+	s := &Server{cfg: cfg}
+	if !cfg.DisableHTTPS {
+		cert, err := generateTLSCert()
+		if err != nil {
+			return nil, err
+		}
+		s.addr = fmt.Sprintf(":%d", cfg.Port)
+		s.presharedKey = cfg.PresharedKey
+		s.mux = http.NewServeMux()
+		s.tlsCert = cert
+		s.registerRoutes()
+	}
+	return s, nil
+}
+
+// Run starts the HTTPS server (unless DisableHTTPS) and blocks until ctx is cancelled or the server errors.
+// When DisableHTTPS is true, Run() blocks on ctx only and does not listen; use ProcessConnection for connection events.
+func (s *Server) Run(ctx context.Context) error {
+	if s.cfg.DisableHTTPS {
+		logger.Debug("auth-daemon running (HTTPS disabled)")
+		<-ctx.Done()
+		s.cleanupPrincipalsFile()
+		return nil
+	}
+	tcfg := &tls.Config{
+		Certificates: []tls.Certificate{s.tlsCert},
+		MinVersion:   tls.VersionTLS12,
+	}
+	handler := s.authMiddleware(s.mux)
+	srv := &http.Server{
+		Addr:              s.addr,
+		Handler:           handler,
+		TLSConfig:         tcfg,
+		ReadTimeout:       10 * time.Second,
+		WriteTimeout:      10 * time.Second,
+		ReadHeaderTimeout: 5 * time.Second,
+		IdleTimeout:       60 * time.Second,
+	}
+	go func() {
+		<-ctx.Done()
+		shutdownCtx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
+		defer cancel()
+		if err := srv.Shutdown(shutdownCtx); err != nil {
+			logger.Warn("auth-daemon shutdown: %v", err)
+		}
+	}()
+	logger.Info("auth-daemon listening on https://127.0.0.1%s", s.addr)
+	if err := srv.ListenAndServeTLS("", ""); err != nil && err != http.ErrServerClosed {
+		return err
+	}
+	s.cleanupPrincipalsFile()
+	return nil
+}
+
+func (s *Server) cleanupPrincipalsFile() {
+	if s.cfg.PrincipalsFilePath != "" {
+		if err := os.Remove(s.cfg.PrincipalsFilePath); err != nil && !os.IsNotExist(err) {
+			logger.Warn("auth-daemon: remove principals file: %v", err)
+		}
+	}
+}

bind/shared_bind.go

@@ -0,0 +1,840 @@
+//go:build !js
+
+package bind
+
+import (
+	"bytes"
+	"fmt"
+	"net"
+	"net/netip"
+	"runtime"
+	"sync"
+	"sync/atomic"
+
+	"github.com/fosrl/newt/logger"
+	"golang.org/x/net/ipv4"
+	"golang.org/x/net/ipv6"
+	wgConn "golang.zx2c4.com/wireguard/conn"
+)
+
+// Magic packet constants for connection testing
+// These packets are intercepted by SharedBind and responded to directly,
+// without being passed to the WireGuard device.
+var (
+	// MagicTestRequest is the prefix for a test request packet
+	// Format: PANGOLIN_TEST_REQ + 8 bytes of random data (for echo)
+	MagicTestRequest = []byte("PANGOLIN_TEST_REQ")
+
+	// MagicTestResponse is the prefix for a test response packet
+	// Format: PANGOLIN_TEST_RSP + 8 bytes echoed from request
+	MagicTestResponse = []byte("PANGOLIN_TEST_RSP")
+)
+
+const (
+	// MagicPacketDataLen is the length of random data included in test packets
+	MagicPacketDataLen = 8
+
+	// MagicTestRequestLen is the total length of a test request packet
+	MagicTestRequestLen = 17 + MagicPacketDataLen // len("PANGOLIN_TEST_REQ") + 8
+
+	// MagicTestResponseLen is the total length of a test response packet
+	MagicTestResponseLen = 17 + MagicPacketDataLen // len("PANGOLIN_TEST_RSP") + 8
+)
+
+// PacketSource identifies where a packet came from
+type PacketSource uint8
+
+const (
+	SourceSocket   PacketSource = iota // From physical UDP socket (hole-punched clients)
+	SourceNetstack                     // From netstack (relay through main tunnel)
+)
+
+// SourceAwareEndpoint wraps an endpoint with source information
+type SourceAwareEndpoint struct {
+	wgConn.Endpoint
+	source PacketSource
+}
+
+// GetSource returns the source of this endpoint
+func (e *SourceAwareEndpoint) GetSource() PacketSource {
+	return e.source
+}
+
+// injectedPacket represents a packet injected into the SharedBind from an internal source
+type injectedPacket struct {
+	data     []byte
+	endpoint wgConn.Endpoint
+}
+
+// Endpoint represents a network endpoint for the SharedBind
+type Endpoint struct {
+	AddrPort netip.AddrPort
+}
+
+// ClearSrc implements the wgConn.Endpoint interface
+func (e *Endpoint) ClearSrc() {}
+
+// DstIP implements the wgConn.Endpoint interface
+func (e *Endpoint) DstIP() netip.Addr {
+	return e.AddrPort.Addr()
+}
+
+// SrcIP implements the wgConn.Endpoint interface
+func (e *Endpoint) SrcIP() netip.Addr {
+	return netip.Addr{}
+}
+
+// DstToBytes implements the wgConn.Endpoint interface
+func (e *Endpoint) DstToBytes() []byte {
+	b, _ := e.AddrPort.MarshalBinary()
+	return b
+}
+
+// DstToString implements the wgConn.Endpoint interface
+func (e *Endpoint) DstToString() string {
+	return e.AddrPort.String()
+}
+
+// SrcToString implements the wgConn.Endpoint interface
+func (e *Endpoint) SrcToString() string {
+	return ""
+}
+
+// SharedBind is a thread-safe UDP bind that can be shared between WireGuard
+// and hole punch senders. It wraps a single UDP connection and implements
+// reference counting to prevent premature closure.
+// It also supports receiving packets from a netstack and routing responses
+// back through the appropriate source.
+type SharedBind struct {
+	mu sync.RWMutex
+
+	// The underlying UDP connection (for hole-punched clients)
+	udpConn *net.UDPConn
+
+	// IPv4 and IPv6 packet connections for advanced features
+	ipv4PC *ipv4.PacketConn
+	ipv6PC *ipv6.PacketConn
+
+	// Reference counting to prevent closing while in use
+	refCount atomic.Int32
+	closed   atomic.Bool
+
+	// Channels for receiving data
+	recvFuncs []wgConn.ReceiveFunc
+
+	// Port binding information
+	port uint16
+
+	// Channel for packets from netstack (from direct relay) - larger buffer for throughput
+	netstackPackets chan injectedPacket
+
+	// Netstack connection for sending responses back through the tunnel
+	// Using atomic.Pointer for lock-free access in hot path
+	netstackConn atomic.Pointer[net.PacketConn]
+
+	// Track which endpoints came from netstack (key: netip.AddrPort, value: struct{})
+	// Using netip.AddrPort directly as key is more efficient than string
+	netstackEndpoints sync.Map
+
+	// Pre-allocated message buffers for batch operations (Linux only)
+	ipv4Msgs []ipv4.Message
+
+	// Shutdown signal for receive goroutines
+	closeChan chan struct{}
+
+	// Callback for magic test responses (used for holepunch testing)
+	magicResponseCallback atomic.Pointer[func(addr netip.AddrPort, echoData []byte)]
+
+	// Rebinding state - used to keep receive goroutines alive during socket transition
+	rebinding     bool       // true when socket is being replaced
+	rebindingCond *sync.Cond // signaled when rebind completes
+}
+
+// MagicResponseCallback is the function signature for magic packet response callbacks
+type MagicResponseCallback func(addr netip.AddrPort, echoData []byte)
+
+// New creates a new SharedBind from an existing UDP connection.
+// The SharedBind takes ownership of the connection and will close it
+// when all references are released.
+func New(udpConn *net.UDPConn) (*SharedBind, error) {
+	if udpConn == nil {
+		return nil, fmt.Errorf("udpConn cannot be nil")
+	}
+
+	bind := &SharedBind{
+		udpConn:         udpConn,
+		netstackPackets: make(chan injectedPacket, 1024), // Larger buffer for better throughput
+		closeChan:       make(chan struct{}),
+	}
+
+	// Initialize the rebinding condition variable
+	bind.rebindingCond = sync.NewCond(&bind.mu)
+
+	// Initialize reference count to 1 (the creator holds the first reference)
+	bind.refCount.Store(1)
+
+	// Get the local port
+	if addr, ok := udpConn.LocalAddr().(*net.UDPAddr); ok {
+		bind.port = uint16(addr.Port)
+	}
+
+	return bind, nil
+}
+
+// SetNetstackConn sets the netstack connection for receiving/sending packets through the tunnel.
+// This connection is used for relay traffic that should go back through the main tunnel.
+func (b *SharedBind) SetNetstackConn(conn net.PacketConn) {
+	b.netstackConn.Store(&conn)
+}
+
+// GetNetstackConn returns the netstack connection if set
+func (b *SharedBind) GetNetstackConn() net.PacketConn {
+	ptr := b.netstackConn.Load()
+	if ptr == nil {
+		return nil
+	}
+	return *ptr
+}
+
+// InjectPacket allows injecting a packet directly into the SharedBind's receive path.
+// This is used for direct relay from netstack without going through the host network.
+// The fromAddr should be the address the packet appears to come from.
+func (b *SharedBind) InjectPacket(data []byte, fromAddr netip.AddrPort) error {
+	if b.closed.Load() {
+		return net.ErrClosed
+	}
+
+	// Unmap IPv4-in-IPv6 addresses to ensure consistency with parsed endpoints
+	if fromAddr.Addr().Is4In6() {
+		fromAddr = netip.AddrPortFrom(fromAddr.Addr().Unmap(), fromAddr.Port())
+	}
+
+	// Track this endpoint as coming from netstack so responses go back the same way
+	// Use AddrPort directly as key (more efficient than string)
+	b.netstackEndpoints.Store(fromAddr, struct{}{})
+
+	// Make a copy of the data to avoid issues with buffer reuse
+	dataCopy := make([]byte, len(data))
+	copy(dataCopy, data)
+
+	select {
+	case b.netstackPackets <- injectedPacket{
+		data:     dataCopy,
+		endpoint: &wgConn.StdNetEndpoint{AddrPort: fromAddr},
+	}:
+		return nil
+	case <-b.closeChan:
+		return net.ErrClosed
+	default:
+		// Channel full, drop the packet
+		return fmt.Errorf("netstack packet buffer full")
+	}
+}
+
+// AddRef increments the reference count. Call this when sharing
+// the bind with another component.
+func (b *SharedBind) AddRef() {
+	newCount := b.refCount.Add(1)
+	// Optional: Add logging for debugging
+	_ = newCount // Placeholder for potential logging
+}
+
+// Release decrements the reference count. When it reaches zero,
+// the underlying UDP connection is closed.
+func (b *SharedBind) Release() error {
+	newCount := b.refCount.Add(-1)
+	// Optional: Add logging for debugging
+	_ = newCount // Placeholder for potential logging
+
+	if newCount < 0 {
+		// This should never happen with proper usage
+		b.refCount.Store(0)
+		return fmt.Errorf("SharedBind reference count went negative")
+	}
+
+	if newCount == 0 {
+		return b.closeConnection()
+	}
+
+	return nil
+}
+
+// closeConnection actually closes the UDP connection
+func (b *SharedBind) closeConnection() error {
+	if !b.closed.CompareAndSwap(false, true) {
+		// Already closed
+		return nil
+	}
+
+	// Signal all goroutines to stop
+	close(b.closeChan)
+
+	b.mu.Lock()
+	defer b.mu.Unlock()
+
+	var err error
+	if b.udpConn != nil {
+		err = b.udpConn.Close()
+		b.udpConn = nil
+	}
+
+	b.ipv4PC = nil
+	b.ipv6PC = nil
+
+	// Clear netstack connection (but don't close it - it's managed externally)
+	b.netstackConn.Store(nil)
+
+	// Clear tracked netstack endpoints
+	b.netstackEndpoints = sync.Map{}
+
+	return err
+}
+
+// ClearNetstackConn clears the netstack connection and tracked endpoints.
+// Call this when stopping the relay.
+func (b *SharedBind) ClearNetstackConn() {
+	b.netstackConn.Store(nil)
+
+	// Clear tracked netstack endpoints
+	b.netstackEndpoints = sync.Map{}
+}
+
+// GetUDPConn returns the underlying UDP connection.
+// The caller must not close this connection directly.
+func (b *SharedBind) GetUDPConn() *net.UDPConn {
+	b.mu.RLock()
+	defer b.mu.RUnlock()
+	return b.udpConn
+}
+
+// GetRefCount returns the current reference count (for debugging)
+func (b *SharedBind) GetRefCount() int32 {
+	return b.refCount.Load()
+}
+
+// IsClosed returns whether the bind is closed
+func (b *SharedBind) IsClosed() bool {
+	return b.closed.Load()
+}
+
+// GetPort returns the current UDP port the bind is using.
+// This is useful when rebinding to try to reuse the same port.
+func (b *SharedBind) GetPort() uint16 {
+	b.mu.RLock()
+	defer b.mu.RUnlock()
+	return b.port
+}
+
+// CloseSocket closes the underlying UDP connection to release the port,
+// but keeps the SharedBind in a state where it can accept a new connection via Rebind.
+// This allows the caller to close the old socket first, then bind a new socket
+// to the same port before calling Rebind.
+//
+// Returns the port that was being used, so the caller can attempt to rebind to it.
+// Sets the rebinding flag so receive goroutines will wait for the new socket
+// instead of exiting.
+func (b *SharedBind) CloseSocket() (uint16, error) {
+	b.mu.Lock()
+	defer b.mu.Unlock()
+
+	if b.closed.Load() {
+		return 0, fmt.Errorf("bind is closed")
+	}
+
+	port := b.port
+
+	// Set rebinding flag BEFORE closing the socket so receive goroutines
+	// know to wait instead of exit
+	b.rebinding = true
+
+	// Close the old connection to release the port
+	if b.udpConn != nil {
+		logger.Debug("Closing UDP connection to release port %d (rebinding)", port)
+		b.udpConn.Close()
+		b.udpConn = nil
+	}
+
+	return port, nil
+}
+
+// Rebind replaces the underlying UDP connection with a new one.
+// This is necessary when network connectivity changes (e.g., WiFi to cellular
+// transition on macOS/iOS) and the old socket becomes stale.
+//
+// The caller is responsible for creating the new UDP connection and passing it here.
+// After rebind, the caller should trigger a hole punch to re-establish NAT mappings.
+//
+// Note: Call CloseSocket() first if you need to rebind to the same port, as the
+// old socket must be closed before a new socket can bind to the same port.
+func (b *SharedBind) Rebind(newConn *net.UDPConn) error {
+	if newConn == nil {
+		return fmt.Errorf("newConn cannot be nil")
+	}
+
+	b.mu.Lock()
+	defer b.mu.Unlock()
+
+	if b.closed.Load() {
+		return fmt.Errorf("bind is closed")
+	}
+
+	// Close the old connection if it's still open
+	// (it may have already been closed via CloseSocket)
+	if b.udpConn != nil {
+		logger.Debug("Closing old UDP connection during rebind")
+		b.udpConn.Close()
+	}
+
+	// Set up the new connection
+	b.udpConn = newConn
+
+	// Update packet connections for the new socket
+	if runtime.GOOS == "linux" || runtime.GOOS == "android" {
+		b.ipv4PC = ipv4.NewPacketConn(newConn)
+		b.ipv6PC = ipv6.NewPacketConn(newConn)
+
+		// Re-initialize message buffers for batch operations
+		batchSize := wgConn.IdealBatchSize
+		b.ipv4Msgs = make([]ipv4.Message, batchSize)
+		for i := range b.ipv4Msgs {
+			b.ipv4Msgs[i].OOB = make([]byte, 0)
+		}
+	} else {
+		// For non-Linux platforms, still set up ipv4PC for consistency
+		b.ipv4PC = ipv4.NewPacketConn(newConn)
+		b.ipv6PC = ipv6.NewPacketConn(newConn)
+	}
+
+	// Update the port
+	if addr, ok := newConn.LocalAddr().(*net.UDPAddr); ok {
+		b.port = uint16(addr.Port)
+		logger.Info("Rebound UDP socket to port %d", b.port)
+	}
+
+	// Clear the rebinding flag and wake up any waiting receive goroutines
+	b.rebinding = false
+	b.rebindingCond.Broadcast()
+
+	logger.Debug("Rebind complete, signaled waiting receive goroutines")
+
+	return nil
+}
+
+// SetMagicResponseCallback sets a callback function that will be called when
+// a magic test response packet is received. This is used for holepunch testing.
+// Pass nil to clear the callback.
+func (b *SharedBind) SetMagicResponseCallback(callback MagicResponseCallback) {
+	if callback == nil {
+		b.magicResponseCallback.Store(nil)
+	} else {
+		// Convert to the function type the atomic.Pointer expects
+		fn := func(addr netip.AddrPort, echoData []byte) {
+			callback(addr, echoData)
+		}
+		b.magicResponseCallback.Store(&fn)
+	}
+}
+
+// WriteToUDP writes data to a specific UDP address.
+// This is thread-safe and can be used by hole punch senders.
+func (b *SharedBind) WriteToUDP(data []byte, addr *net.UDPAddr) (int, error) {
+	if b.closed.Load() {
+		return 0, net.ErrClosed
+	}
+
+	b.mu.RLock()
+	conn := b.udpConn
+	b.mu.RUnlock()
+
+	if conn == nil {
+		return 0, net.ErrClosed
+	}
+
+	return conn.WriteToUDP(data, addr)
+}
+
+// Close implements the WireGuard Bind interface.
+// It decrements the reference count and closes the connection if no references remain.
+func (b *SharedBind) Close() error {
+	return b.Release()
+}
+
+// Open implements the WireGuard Bind interface.
+// Since the connection is already open, this just sets up the receive functions.
+func (b *SharedBind) Open(uport uint16) ([]wgConn.ReceiveFunc, uint16, error) {
+	if b.closed.Load() {
+		return nil, 0, net.ErrClosed
+	}
+
+	b.mu.Lock()
+	defer b.mu.Unlock()
+
+	if b.udpConn == nil {
+		return nil, 0, net.ErrClosed
+	}
+
+	// Set up IPv4 and IPv6 packet connections for advanced features
+	if runtime.GOOS == "linux" || runtime.GOOS == "android" {
+		b.ipv4PC = ipv4.NewPacketConn(b.udpConn)
+		b.ipv6PC = ipv6.NewPacketConn(b.udpConn)
+
+		// Pre-allocate message buffers for batch operations
+		batchSize := wgConn.IdealBatchSize
+		b.ipv4Msgs = make([]ipv4.Message, batchSize)
+		for i := range b.ipv4Msgs {
+			b.ipv4Msgs[i].OOB = make([]byte, 0)
+		}
+	}
+
+	// Create receive functions - one for socket, one for netstack
+	recvFuncs := make([]wgConn.ReceiveFunc, 0, 2)
+
+	// Add socket receive function (reads from physical UDP socket)
+	recvFuncs = append(recvFuncs, b.makeReceiveSocket())
+
+	// Add netstack receive function (reads from injected packets channel)
+	recvFuncs = append(recvFuncs, b.makeReceiveNetstack())
+
+	b.recvFuncs = recvFuncs
+	return recvFuncs, b.port, nil
+}
+
+// makeReceiveSocket creates a receive function for physical UDP socket packets
+func (b *SharedBind) makeReceiveSocket() wgConn.ReceiveFunc {
+	return func(bufs [][]byte, sizes []int, eps []wgConn.Endpoint) (n int, err error) {
+		for {
+			if b.closed.Load() {
+				return 0, net.ErrClosed
+			}
+
+			b.mu.RLock()
+			conn := b.udpConn
+			pc := b.ipv4PC
+			b.mu.RUnlock()
+
+			if conn == nil {
+				// Socket is nil - check if we're rebinding or truly closed
+				if b.closed.Load() {
+					return 0, net.ErrClosed
+				}
+
+				// Wait for rebind to complete
+				b.mu.Lock()
+				for b.rebinding && !b.closed.Load() {
+					logger.Debug("Receive goroutine waiting for socket rebind to complete")
+					b.rebindingCond.Wait()
+				}
+				b.mu.Unlock()
+
+				// Check again after waking up
+				if b.closed.Load() {
+					return 0, net.ErrClosed
+				}
+
+				// Loop back to retry with new socket
+				continue
+			}
+
+			// Use batch reading on Linux for performance
+			var n int
+			var err error
+			if pc != nil && (runtime.GOOS == "linux" || runtime.GOOS == "android") {
+				n, err = b.receiveIPv4Batch(pc, bufs, sizes, eps)
+			} else {
+				n, err = b.receiveIPv4Simple(conn, bufs, sizes, eps)
+			}
+
+			if err != nil {
+				// Check if this error is due to rebinding
+				b.mu.RLock()
+				rebinding := b.rebinding
+				b.mu.RUnlock()
+
+				if rebinding {
+					logger.Debug("Receive got error during rebind, waiting for new socket: %v", err)
+					// Wait for rebind to complete and retry
+					b.mu.Lock()
+					for b.rebinding && !b.closed.Load() {
+						b.rebindingCond.Wait()
+					}
+					b.mu.Unlock()
+
+					if b.closed.Load() {
+						return 0, net.ErrClosed
+					}
+
+					// Retry with new socket
+					continue
+				}
+
+				// Not rebinding, return the error
+				return 0, err
+			}
+
+			return n, nil
+		}
+	}
+}
+
+// makeReceiveNetstack creates a receive function for netstack-injected packets
+func (b *SharedBind) makeReceiveNetstack() wgConn.ReceiveFunc {
+	return func(bufs [][]byte, sizes []int, eps []wgConn.Endpoint) (n int, err error) {
+		select {
+		case <-b.closeChan:
+			return 0, net.ErrClosed
+		case pkt := <-b.netstackPackets:
+			if len(pkt.data) <= len(bufs[0]) {
+				copy(bufs[0], pkt.data)
+				sizes[0] = len(pkt.data)
+				eps[0] = pkt.endpoint
+				return 1, nil
+			}
+			// Packet too large for buffer, skip it
+			return 0, nil
+		}
+	}
+}
+
+// receiveIPv4Batch uses batch reading for better performance on Linux
+func (b *SharedBind) receiveIPv4Batch(pc *ipv4.PacketConn, bufs [][]byte, sizes []int, eps []wgConn.Endpoint) (int, error) {
+	// Use pre-allocated messages, just update buffer pointers
+	numBufs := len(bufs)
+	if numBufs > len(b.ipv4Msgs) {
+		numBufs = len(b.ipv4Msgs)
+	}
+
+	for i := 0; i < numBufs; i++ {
+		b.ipv4Msgs[i].Buffers = [][]byte{bufs[i]}
+	}
+
+	numMsgs, err := pc.ReadBatch(b.ipv4Msgs[:numBufs], 0)
+	if err != nil {
+		return 0, err
+	}
+
+	// Process messages and filter out magic packets
+	writeIdx := 0
+	for i := 0; i < numMsgs; i++ {
+		if b.ipv4Msgs[i].N == 0 {
+			continue
+		}
+
+		// Check for magic packet
+		if b.ipv4Msgs[i].Addr != nil {
+			if udpAddr, ok := b.ipv4Msgs[i].Addr.(*net.UDPAddr); ok {
+				data := bufs[i][:b.ipv4Msgs[i].N]
+				if b.handleMagicPacket(data, udpAddr) {
+					// Magic packet handled, skip this message
+					continue
+				}
+			}
+		}
+
+		// Not a magic packet, include in output
+		if writeIdx != i {
+			// Need to copy data to the correct position
+			copy(bufs[writeIdx], bufs[i][:b.ipv4Msgs[i].N])
+		}
+		sizes[writeIdx] = b.ipv4Msgs[i].N
+
+		if b.ipv4Msgs[i].Addr != nil {
+			if udpAddr, ok := b.ipv4Msgs[i].Addr.(*net.UDPAddr); ok {
+				addrPort := udpAddr.AddrPort()
+				// Unmap IPv4-in-IPv6 addresses to ensure consistency with parsed endpoints
+				if addrPort.Addr().Is4In6() {
+					addrPort = netip.AddrPortFrom(addrPort.Addr().Unmap(), addrPort.Port())
+				}
+				eps[writeIdx] = &wgConn.StdNetEndpoint{AddrPort: addrPort}
+			}
+		}
+		writeIdx++
+	}
+
+	return writeIdx, nil
+}
+
+// receiveIPv4Simple uses simple ReadFromUDP for non-Linux platforms
+func (b *SharedBind) receiveIPv4Simple(conn *net.UDPConn, bufs [][]byte, sizes []int, eps []wgConn.Endpoint) (int, error) {
+	// No read deadline - we rely on socket close to unblock during rebind.
+	// The caller (makeReceiveSocket) handles rebind state when errors occur.
+	for {
+		n, addr, err := conn.ReadFromUDP(bufs[0])
+		if err != nil {
+			return 0, err
+		}
+
+		// Check for magic test packet and handle it directly
+		if b.handleMagicPacket(bufs[0][:n], addr) {
+			// Magic packet was handled, read another packet
+			continue
+		}
+
+		sizes[0] = n
+		if addr != nil {
+			addrPort := addr.AddrPort()
+			// Unmap IPv4-in-IPv6 addresses to ensure consistency with parsed endpoints
+			if addrPort.Addr().Is4In6() {
+				addrPort = netip.AddrPortFrom(addrPort.Addr().Unmap(), addrPort.Port())
+			}
+			eps[0] = &wgConn.StdNetEndpoint{AddrPort: addrPort}
+		}
+
+		return 1, nil
+	}
+}
+
+// handleMagicPacket checks if the packet is a magic test packet and responds if so.
+// Returns true if the packet was a magic packet and was handled (should not be passed to WireGuard).
+func (b *SharedBind) handleMagicPacket(data []byte, addr *net.UDPAddr) bool {
+	// Check if this is a test request packet
+	if len(data) >= MagicTestRequestLen && bytes.HasPrefix(data, MagicTestRequest) {
+		// logger.Debug("Received magic test REQUEST from %s, sending response", addr.String())
+		// Extract the random data portion to echo back
+		echoData := data[len(MagicTestRequest) : len(MagicTestRequest)+MagicPacketDataLen]
+
+		// Build response packet
+		response := make([]byte, MagicTestResponseLen)
+		copy(response, MagicTestResponse)
+		copy(response[len(MagicTestResponse):], echoData)
+
+		// Send response back to sender
+		b.mu.RLock()
+		conn := b.udpConn
+		b.mu.RUnlock()
+
+		if conn != nil {
+			_, _ = conn.WriteToUDP(response, addr)
+		}
+
+		return true
+	}
+
+	// Check if this is a test response packet
+	if len(data) >= MagicTestResponseLen && bytes.HasPrefix(data, MagicTestResponse) {
+		// logger.Debug("Received magic test RESPONSE from %s", addr.String())
+		// Extract the echoed data
+		echoData := data[len(MagicTestResponse) : len(MagicTestResponse)+MagicPacketDataLen]
+
+		// Call the callback if set
+		callbackPtr := b.magicResponseCallback.Load()
+		if callbackPtr != nil {
+			callback := *callbackPtr
+			addrPort := addr.AddrPort()
+			// Unmap IPv4-in-IPv6 addresses to ensure consistency
+			if addrPort.Addr().Is4In6() {
+				addrPort = netip.AddrPortFrom(addrPort.Addr().Unmap(), addrPort.Port())
+			}
+			callback(addrPort, echoData)
+		} else {
+			logger.Debug("Magic response received but no callback registered")
+		}
+
+		return true
+	}
+
+	return false
+}
+
+// Send implements the WireGuard Bind interface.
+// It sends packets to the specified endpoint, routing through the appropriate
+// source (netstack or physical socket) based on where the endpoint's packets came from.
+func (b *SharedBind) Send(bufs [][]byte, ep wgConn.Endpoint) error {
+	if b.closed.Load() {
+		return net.ErrClosed
+	}
+
+	// Extract the destination address from the endpoint
+	var destAddrPort netip.AddrPort
+
+	// Try to cast to StdNetEndpoint first (most common case, avoid allocations)
+	if stdEp, ok := ep.(*wgConn.StdNetEndpoint); ok {
+		destAddrPort = stdEp.AddrPort
+	} else {
+		// Fallback: construct from DstIP and DstToBytes
+		dstBytes := ep.DstToBytes()
+		if len(dstBytes) >= 6 { // Minimum for IPv4 (4 bytes) + port (2 bytes)
+			var addr netip.Addr
+			var port uint16
+
+			if len(dstBytes) >= 18 { // IPv6 (16 bytes) + port (2 bytes)
+				addr, _ = netip.AddrFromSlice(dstBytes[:16])
+				port = uint16(dstBytes[16]) | uint16(dstBytes[17])<<8
+			} else { // IPv4
+				addr, _ = netip.AddrFromSlice(dstBytes[:4])
+				port = uint16(dstBytes[4]) | uint16(dstBytes[5])<<8
+			}
+
+			if addr.IsValid() {
+				destAddrPort = netip.AddrPortFrom(addr, port)
+			}
+		}
+	}
+
+	if !destAddrPort.IsValid() {
+		return fmt.Errorf("could not extract destination address from endpoint")
+	}
+
+	// Check if this endpoint came from netstack - if so, send through netstack
+	// Use AddrPort directly as key (more efficient than string conversion)
+	if _, isNetstackEndpoint := b.netstackEndpoints.Load(destAddrPort); isNetstackEndpoint {
+		connPtr := b.netstackConn.Load()
+		if connPtr != nil && *connPtr != nil {
+			netstackConn := *connPtr
+			destAddr := net.UDPAddrFromAddrPort(destAddrPort)
+			// Send all buffers through netstack
+			for _, buf := range bufs {
+				_, err := netstackConn.WriteTo(buf, destAddr)
+				if err != nil {
+					return err
+				}
+			}
+			return nil
+		}
+		// Fall through to socket if netstack conn not available
+	}
+
+	// Send through the physical UDP socket (for hole-punched clients)
+	b.mu.RLock()
+	conn := b.udpConn
+	b.mu.RUnlock()
+
+	if conn == nil {
+		return net.ErrClosed
+	}
+
+	destAddr := net.UDPAddrFromAddrPort(destAddrPort)
+
+	// Send all buffers to the destination
+	for _, buf := range bufs {
+		_, err := conn.WriteToUDP(buf, destAddr)
+		if err != nil {
+			return err
+		}
+	}
+
+	return nil
+}
+
+// SetMark implements the WireGuard Bind interface.
+// It's a no-op for this implementation.
+func (b *SharedBind) SetMark(mark uint32) error {
+	// Not implemented for this use case
+	return nil
+}
+
+// BatchSize returns the preferred batch size for sending packets.
+func (b *SharedBind) BatchSize() int {
+	if runtime.GOOS == "linux" || runtime.GOOS == "android" {
+		return wgConn.IdealBatchSize
+	}
+	return 1
+}
+
+// ParseEndpoint creates a new endpoint from a string address.
+func (b *SharedBind) ParseEndpoint(s string) (wgConn.Endpoint, error) {
+	addrPort, err := netip.ParseAddrPort(s)
+	if err != nil {
+		return nil, err
+	}
+	return &wgConn.StdNetEndpoint{AddrPort: addrPort}, nil
+}

bind/shared_bind_test.go

@@ -0,0 +1,555 @@
+//go:build !js
+
+package bind
+
+import (
+	"net"
+	"net/netip"
+	"sync"
+	"testing"
+	"time"
+
+	wgConn "golang.zx2c4.com/wireguard/conn"
+)
+
+// TestSharedBindCreation tests basic creation and initialization
+func TestSharedBindCreation(t *testing.T) {
+	// Create a UDP connection
+	udpConn, err := net.ListenUDP("udp", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		t.Fatalf("Failed to create UDP connection: %v", err)
+	}
+	defer udpConn.Close()
+
+	// Create SharedBind
+	bind, err := New(udpConn)
+	if err != nil {
+		t.Fatalf("Failed to create SharedBind: %v", err)
+	}
+
+	if bind == nil {
+		t.Fatal("SharedBind is nil")
+	}
+
+	// Verify initial reference count
+	if bind.refCount.Load() != 1 {
+		t.Errorf("Expected initial refCount to be 1, got %d", bind.refCount.Load())
+	}
+
+	// Clean up
+	if err := bind.Close(); err != nil {
+		t.Errorf("Failed to close SharedBind: %v", err)
+	}
+}
+
+// TestSharedBindReferenceCount tests reference counting
+func TestSharedBindReferenceCount(t *testing.T) {
+	udpConn, err := net.ListenUDP("udp", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		t.Fatalf("Failed to create UDP connection: %v", err)
+	}
+
+	bind, err := New(udpConn)
+	if err != nil {
+		t.Fatalf("Failed to create SharedBind: %v", err)
+	}
+
+	// Add references
+	bind.AddRef()
+	if bind.refCount.Load() != 2 {
+		t.Errorf("Expected refCount to be 2, got %d", bind.refCount.Load())
+	}
+
+	bind.AddRef()
+	if bind.refCount.Load() != 3 {
+		t.Errorf("Expected refCount to be 3, got %d", bind.refCount.Load())
+	}
+
+	// Release references
+	bind.Release()
+	if bind.refCount.Load() != 2 {
+		t.Errorf("Expected refCount to be 2 after release, got %d", bind.refCount.Load())
+	}
+
+	bind.Release()
+	bind.Release() // This should close the connection
+
+	if !bind.closed.Load() {
+		t.Error("Expected bind to be closed after all references released")
+	}
+}
+
+// TestSharedBindWriteToUDP tests the WriteToUDP functionality
+func TestSharedBindWriteToUDP(t *testing.T) {
+	// Create sender
+	senderConn, err := net.ListenUDP("udp", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		t.Fatalf("Failed to create sender UDP connection: %v", err)
+	}
+
+	senderBind, err := New(senderConn)
+	if err != nil {
+		t.Fatalf("Failed to create sender SharedBind: %v", err)
+	}
+	defer senderBind.Close()
+
+	// Create receiver
+	receiverConn, err := net.ListenUDP("udp", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		t.Fatalf("Failed to create receiver UDP connection: %v", err)
+	}
+	defer receiverConn.Close()
+
+	receiverAddr := receiverConn.LocalAddr().(*net.UDPAddr)
+
+	// Send data
+	testData := []byte("Hello, SharedBind!")
+	n, err := senderBind.WriteToUDP(testData, receiverAddr)
+	if err != nil {
+		t.Fatalf("WriteToUDP failed: %v", err)
+	}
+
+	if n != len(testData) {
+		t.Errorf("Expected to send %d bytes, sent %d", len(testData), n)
+	}
+
+	// Receive data
+	buf := make([]byte, 1024)
+	receiverConn.SetReadDeadline(time.Now().Add(2 * time.Second))
+	n, _, err = receiverConn.ReadFromUDP(buf)
+	if err != nil {
+		t.Fatalf("Failed to receive data: %v", err)
+	}
+
+	if string(buf[:n]) != string(testData) {
+		t.Errorf("Expected to receive %q, got %q", testData, buf[:n])
+	}
+}
+
+// TestSharedBindConcurrentWrites tests thread-safety
+func TestSharedBindConcurrentWrites(t *testing.T) {
+	// Create sender
+	senderConn, err := net.ListenUDP("udp", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		t.Fatalf("Failed to create sender UDP connection: %v", err)
+	}
+
+	senderBind, err := New(senderConn)
+	if err != nil {
+		t.Fatalf("Failed to create sender SharedBind: %v", err)
+	}
+	defer senderBind.Close()
+
+	// Create receiver
+	receiverConn, err := net.ListenUDP("udp", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		t.Fatalf("Failed to create receiver UDP connection: %v", err)
+	}
+	defer receiverConn.Close()
+
+	receiverAddr := receiverConn.LocalAddr().(*net.UDPAddr)
+
+	// Launch concurrent writes
+	numGoroutines := 100
+	var wg sync.WaitGroup
+	wg.Add(numGoroutines)
+
+	for i := 0; i < numGoroutines; i++ {
+		go func(id int) {
+			defer wg.Done()
+			data := []byte{byte(id)}
+			_, err := senderBind.WriteToUDP(data, receiverAddr)
+			if err != nil {
+				t.Errorf("WriteToUDP failed in goroutine %d: %v", id, err)
+			}
+		}(i)
+	}
+
+	wg.Wait()
+}
+
+// TestSharedBindWireGuardInterface tests WireGuard Bind interface implementation
+func TestSharedBindWireGuardInterface(t *testing.T) {
+	udpConn, err := net.ListenUDP("udp", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		t.Fatalf("Failed to create UDP connection: %v", err)
+	}
+
+	bind, err := New(udpConn)
+	if err != nil {
+		t.Fatalf("Failed to create SharedBind: %v", err)
+	}
+	defer bind.Close()
+
+	// Test Open
+	recvFuncs, port, err := bind.Open(0)
+	if err != nil {
+		t.Fatalf("Open failed: %v", err)
+	}
+
+	if len(recvFuncs) == 0 {
+		t.Error("Expected at least one receive function")
+	}
+
+	if port == 0 {
+		t.Error("Expected non-zero port")
+	}
+
+	// Test SetMark (should be a no-op)
+	if err := bind.SetMark(0); err != nil {
+		t.Errorf("SetMark failed: %v", err)
+	}
+
+	// Test BatchSize
+	batchSize := bind.BatchSize()
+	if batchSize <= 0 {
+		t.Error("Expected positive batch size")
+	}
+}
+
+// TestSharedBindSend tests the Send method with WireGuard endpoints
+func TestSharedBindSend(t *testing.T) {
+	// Create sender
+	senderConn, err := net.ListenUDP("udp", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		t.Fatalf("Failed to create sender UDP connection: %v", err)
+	}
+
+	senderBind, err := New(senderConn)
+	if err != nil {
+		t.Fatalf("Failed to create sender SharedBind: %v", err)
+	}
+	defer senderBind.Close()
+
+	// Create receiver
+	receiverConn, err := net.ListenUDP("udp", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		t.Fatalf("Failed to create receiver UDP connection: %v", err)
+	}
+	defer receiverConn.Close()
+
+	receiverAddr := receiverConn.LocalAddr().(*net.UDPAddr)
+
+	// Create an endpoint
+	addrPort := receiverAddr.AddrPort()
+	endpoint := &wgConn.StdNetEndpoint{AddrPort: addrPort}
+
+	// Send data
+	testData := []byte("WireGuard packet")
+	bufs := [][]byte{testData}
+	err = senderBind.Send(bufs, endpoint)
+	if err != nil {
+		t.Fatalf("Send failed: %v", err)
+	}
+
+	// Receive data
+	buf := make([]byte, 1024)
+	receiverConn.SetReadDeadline(time.Now().Add(2 * time.Second))
+	n, _, err := receiverConn.ReadFromUDP(buf)
+	if err != nil {
+		t.Fatalf("Failed to receive data: %v", err)
+	}
+
+	if string(buf[:n]) != string(testData) {
+		t.Errorf("Expected to receive %q, got %q", testData, buf[:n])
+	}
+}
+
+// TestSharedBindMultipleUsers simulates WireGuard and hole punch using the same bind
+func TestSharedBindMultipleUsers(t *testing.T) {
+	// Create shared bind
+	udpConn, err := net.ListenUDP("udp", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		t.Fatalf("Failed to create UDP connection: %v", err)
+	}
+
+	sharedBind, err := New(udpConn)
+	if err != nil {
+		t.Fatalf("Failed to create SharedBind: %v", err)
+	}
+
+	// Add reference for hole punch sender
+	sharedBind.AddRef()
+
+	// Create receiver
+	receiverConn, err := net.ListenUDP("udp", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		t.Fatalf("Failed to create receiver UDP connection: %v", err)
+	}
+	defer receiverConn.Close()
+
+	receiverAddr := receiverConn.LocalAddr().(*net.UDPAddr)
+
+	var wg sync.WaitGroup
+
+	// Simulate WireGuard using the bind
+	wg.Add(1)
+	go func() {
+		defer wg.Done()
+		addrPort := receiverAddr.AddrPort()
+		endpoint := &wgConn.StdNetEndpoint{AddrPort: addrPort}
+
+		for i := 0; i < 10; i++ {
+			data := []byte("WireGuard packet")
+			bufs := [][]byte{data}
+			if err := sharedBind.Send(bufs, endpoint); err != nil {
+				t.Errorf("WireGuard Send failed: %v", err)
+			}
+			time.Sleep(10 * time.Millisecond)
+		}
+	}()
+
+	// Simulate hole punch sender using the bind
+	wg.Add(1)
+	go func() {
+		defer wg.Done()
+		for i := 0; i < 10; i++ {
+			data := []byte("Hole punch packet")
+			if _, err := sharedBind.WriteToUDP(data, receiverAddr); err != nil {
+				t.Errorf("Hole punch WriteToUDP failed: %v", err)
+			}
+			time.Sleep(10 * time.Millisecond)
+		}
+	}()
+
+	wg.Wait()
+
+	// Release the hole punch reference
+	sharedBind.Release()
+
+	// Close WireGuard's reference (should close the connection)
+	sharedBind.Close()
+
+	if !sharedBind.closed.Load() {
+		t.Error("Expected bind to be closed after all users released it")
+	}
+}
+
+// TestEndpoint tests the Endpoint implementation
+func TestEndpoint(t *testing.T) {
+	addr := netip.MustParseAddr("192.168.1.1")
+	addrPort := netip.AddrPortFrom(addr, 51820)
+
+	ep := &Endpoint{AddrPort: addrPort}
+
+	// Test DstIP
+	if ep.DstIP() != addr {
+		t.Errorf("Expected DstIP to be %v, got %v", addr, ep.DstIP())
+	}
+
+	// Test DstToString
+	expected := "192.168.1.1:51820"
+	if ep.DstToString() != expected {
+		t.Errorf("Expected DstToString to be %q, got %q", expected, ep.DstToString())
+	}
+
+	// Test DstToBytes
+	bytes := ep.DstToBytes()
+	if len(bytes) == 0 {
+		t.Error("Expected DstToBytes to return non-empty slice")
+	}
+
+	// Test SrcIP (should be zero)
+	if ep.SrcIP().IsValid() {
+		t.Error("Expected SrcIP to be invalid")
+	}
+
+	// Test ClearSrc (should not panic)
+	ep.ClearSrc()
+}
+
+// TestParseEndpoint tests the ParseEndpoint method
+func TestParseEndpoint(t *testing.T) {
+	udpConn, err := net.ListenUDP("udp", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		t.Fatalf("Failed to create UDP connection: %v", err)
+	}
+
+	bind, err := New(udpConn)
+	if err != nil {
+		t.Fatalf("Failed to create SharedBind: %v", err)
+	}
+	defer bind.Close()
+
+	tests := []struct {
+		name      string
+		input     string
+		wantErr   bool
+		checkAddr func(*testing.T, wgConn.Endpoint)
+	}{
+		{
+			name:    "valid IPv4",
+			input:   "192.168.1.1:51820",
+			wantErr: false,
+			checkAddr: func(t *testing.T, ep wgConn.Endpoint) {
+				if ep.DstToString() != "192.168.1.1:51820" {
+					t.Errorf("Expected 192.168.1.1:51820, got %s", ep.DstToString())
+				}
+			},
+		},
+		{
+			name:    "valid IPv6",
+			input:   "[::1]:51820",
+			wantErr: false,
+			checkAddr: func(t *testing.T, ep wgConn.Endpoint) {
+				if ep.DstToString() != "[::1]:51820" {
+					t.Errorf("Expected [::1]:51820, got %s", ep.DstToString())
+				}
+			},
+		},
+		{
+			name:    "invalid - missing port",
+			input:   "192.168.1.1",
+			wantErr: true,
+		},
+		{
+			name:    "invalid - bad format",
+			input:   "not-an-address",
+			wantErr: true,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			ep, err := bind.ParseEndpoint(tt.input)
+			if (err != nil) != tt.wantErr {
+				t.Errorf("ParseEndpoint() error = %v, wantErr %v", err, tt.wantErr)
+				return
+			}
+			if !tt.wantErr && tt.checkAddr != nil {
+				tt.checkAddr(t, ep)
+			}
+		})
+	}
+}
+
+// TestNetstackRouting tests that packets from netstack endpoints are routed back through netstack
+func TestNetstackRouting(t *testing.T) {
+	// Create the SharedBind with a physical UDP socket
+	physicalConn, err := net.ListenUDP("udp", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		t.Fatalf("Failed to create physical UDP connection: %v", err)
+	}
+
+	sharedBind, err := New(physicalConn)
+	if err != nil {
+		t.Fatalf("Failed to create SharedBind: %v", err)
+	}
+	defer sharedBind.Close()
+
+	// Create a mock "netstack" connection (just another UDP socket for testing)
+	netstackConn, err := net.ListenUDP("udp", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		t.Fatalf("Failed to create netstack UDP connection: %v", err)
+	}
+	defer netstackConn.Close()
+
+	// Set the netstack connection
+	sharedBind.SetNetstackConn(netstackConn)
+
+	// Create a "client" that would receive packets
+	clientConn, err := net.ListenUDP("udp", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		t.Fatalf("Failed to create client UDP connection: %v", err)
+	}
+	defer clientConn.Close()
+
+	clientAddr := clientConn.LocalAddr().(*net.UDPAddr)
+	clientAddrPort := clientAddr.AddrPort()
+
+	// Inject a packet from the "netstack" source - this should track the endpoint
+	testData := []byte("test packet from netstack")
+	err = sharedBind.InjectPacket(testData, clientAddrPort)
+	if err != nil {
+		t.Fatalf("InjectPacket failed: %v", err)
+	}
+
+	// Now when we send a response to this endpoint, it should go through netstack
+	endpoint := &wgConn.StdNetEndpoint{AddrPort: clientAddrPort}
+	responseData := []byte("response packet")
+	err = sharedBind.Send([][]byte{responseData}, endpoint)
+	if err != nil {
+		t.Fatalf("Send failed: %v", err)
+	}
+
+	// The packet should be received by the client from the netstack connection
+	buf := make([]byte, 1024)
+	clientConn.SetReadDeadline(time.Now().Add(2 * time.Second))
+	n, fromAddr, err := clientConn.ReadFromUDP(buf)
+	if err != nil {
+		t.Fatalf("Failed to receive response: %v", err)
+	}
+
+	if string(buf[:n]) != string(responseData) {
+		t.Errorf("Expected to receive %q, got %q", responseData, buf[:n])
+	}
+
+	// Verify the response came from the netstack connection, not the physical one
+	netstackAddr := netstackConn.LocalAddr().(*net.UDPAddr)
+	if fromAddr.Port != netstackAddr.Port {
+		t.Errorf("Expected response from netstack port %d, got %d", netstackAddr.Port, fromAddr.Port)
+	}
+}
+
+// TestSocketRouting tests that packets from socket endpoints are routed through socket
+func TestSocketRouting(t *testing.T) {
+	// Create the SharedBind with a physical UDP socket
+	physicalConn, err := net.ListenUDP("udp", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		t.Fatalf("Failed to create physical UDP connection: %v", err)
+	}
+
+	sharedBind, err := New(physicalConn)
+	if err != nil {
+		t.Fatalf("Failed to create SharedBind: %v", err)
+	}
+	defer sharedBind.Close()
+
+	// Create a mock "netstack" connection
+	netstackConn, err := net.ListenUDP("udp", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		t.Fatalf("Failed to create netstack UDP connection: %v", err)
+	}
+	defer netstackConn.Close()
+
+	// Set the netstack connection
+	sharedBind.SetNetstackConn(netstackConn)
+
+	// Create a "client" that would receive packets (this simulates a hole-punched client)
+	clientConn, err := net.ListenUDP("udp", &net.UDPAddr{IP: net.IPv4zero, Port: 0})
+	if err != nil {
+		t.Fatalf("Failed to create client UDP connection: %v", err)
+	}
+	defer clientConn.Close()
+
+	clientAddr := clientConn.LocalAddr().(*net.UDPAddr)
+	clientAddrPort := clientAddr.AddrPort()
+
+	// Don't inject from netstack - this endpoint is NOT tracked as netstack-sourced
+	// So Send should use the physical socket
+
+	endpoint := &wgConn.StdNetEndpoint{AddrPort: clientAddrPort}
+	responseData := []byte("response packet via socket")
+	err = sharedBind.Send([][]byte{responseData}, endpoint)
+	if err != nil {
+		t.Fatalf("Send failed: %v", err)
+	}
+
+	// The packet should be received by the client from the physical connection
+	buf := make([]byte, 1024)
+	clientConn.SetReadDeadline(time.Now().Add(2 * time.Second))
+	n, fromAddr, err := clientConn.ReadFromUDP(buf)
+	if err != nil {
+		t.Fatalf("Failed to receive response: %v", err)
+	}
+
+	if string(buf[:n]) != string(responseData) {
+		t.Errorf("Expected to receive %q, got %q", responseData, buf[:n])
+	}
+
+	// Verify the response came from the physical connection, not the netstack one
+	physicalAddr := physicalConn.LocalAddr().(*net.UDPAddr)
+	if fromAddr.Port != physicalAddr.Port {
+		t.Errorf("Expected response from physical port %d, got %d", physicalAddr.Port, fromAddr.Port)
+	}
+}

browsergateway/browsergateway.go

@@ -0,0 +1,144 @@
+package browsergateway
+
+import (
+	"crypto/subtle"
+	"errors"
+	"net"
+	"net/http"
+	"strings"
+	"sync"
+
+	"github.com/fosrl/newt/nativessh"
+)
+
+// Forwarding buffer size. RDP graphics traffic is bursty and TLS records cap
+// at ~16 KiB, so 64 KiB lets a couple of records pile up per syscall/frame
+// without wasting memory per session.
+const forwardBufSize = 64 * 1024
+
+// ListenPort is the port the browser gateway HTTP server listens on inside the
+// WireGuard netstack. This is a fixed value shared between newt and pangolin.
+// Targets do not overlap with this port because they start at 40000.
+const ListenPort = 39999
+
+// Target represents an allowed proxy destination for the browser gateway.
+// Only connections whose (Type, Destination, DestinationPort, AuthToken) match a
+// registered Target will be forwarded; all others are rejected.
+type Target struct {
+	ID              int
+	Type            string // "rdp" | "ssh" | "vnc"
+	Destination     string
+	DestinationPort int
+	AuthToken       string // per-target secret; must match the token supplied by the client
+}
+
+// Config holds the configuration for a Gateway.
+type Config struct {
+	// AuthToken is used only for NativeSSH mode (which has no external target
+	// to match against). For all proxy targets (RDP/SSH/VNC), auth tokens are
+	// stored per-Target and validated by isAllowed.
+	AuthToken string
+	// SSHCredentials are used by native SSH browser sessions for certificate
+	// validation against the in-memory CA/principal store.
+	SSHCredentials *nativessh.CredentialStore
+}
+
+// Gateway is a browser-based RDP/SSH/VNC WebSocket proxy.
+// Create one with New and mount it via RegisterHandlers or the individual
+// HandleRDP / HandleSSH / HandleVNC http.HandlerFunc methods.
+type Gateway struct {
+	authToken string
+	sshCreds  *nativessh.CredentialStore
+
+	mu      sync.RWMutex
+	targets map[int]Target // keyed by Target.ID
+
+	server *http.Server
+}
+
+// New creates a new Gateway from the provided Config.
+func New(cfg Config) *Gateway {
+	return &Gateway{
+		authToken: cfg.AuthToken,
+		sshCreds:  cfg.SSHCredentials,
+		targets:   make(map[int]Target),
+	}
+}
+
+// SetTargets replaces the entire allowed-destination list atomically.
+func (g *Gateway) SetTargets(targets []Target) {
+	g.mu.Lock()
+	defer g.mu.Unlock()
+	g.targets = make(map[int]Target, len(targets))
+	for _, t := range targets {
+		g.targets[t.ID] = t
+	}
+}
+
+// AddTarget adds or updates a single allowed destination.
+func (g *Gateway) AddTarget(t Target) {
+	g.mu.Lock()
+	defer g.mu.Unlock()
+	g.targets[t.ID] = t
+}
+
+// RemoveTarget removes an allowed destination by its ID.
+func (g *Gateway) RemoveTarget(id int) {
+	g.mu.Lock()
+	defer g.mu.Unlock()
+	delete(g.targets, id)
+}
+
+// isAllowed reports whether a connection to (targetType, host, port) with the
+// given authToken is permitted. The token is compared against the per-target
+// AuthToken using constant-time comparison to prevent timing attacks.
+func (g *Gateway) isAllowed(targetType, host string, port int, authToken string) bool {
+	g.mu.RLock()
+	defer g.mu.RUnlock()
+	for _, t := range g.targets {
+		if t.Type == targetType && t.Destination == host && t.DestinationPort == port {
+			return subtle.ConstantTimeCompare([]byte(authToken), []byte(t.AuthToken)) == 1
+		}
+	}
+	return false
+}
+
+// isTokenValid reports whether the given authToken matches any registered
+// target of the specified type. Used for native SSH mode where there is no
+// external destination to match against.
+func (g *Gateway) isTokenValid(targetType, authToken string) bool {
+	g.mu.RLock()
+	defer g.mu.RUnlock()
+	for _, t := range g.targets {
+		if t.Type == targetType {
+			if subtle.ConstantTimeCompare([]byte(authToken), []byte(t.AuthToken)) == 1 {
+				return true
+			}
+		}
+	}
+	return false
+}
+
+// Start serves the browser gateway HTTP server on the provided listener.
+// It returns nil when the listener is closed (normal shutdown).
+func (g *Gateway) Start(ln net.Listener) error {
+	mux := http.NewServeMux()
+	g.RegisterHandlers(mux)
+	g.server = &http.Server{Handler: mux}
+	err := g.server.Serve(ln)
+	if err == nil ||
+		errors.Is(err, net.ErrClosed) ||
+		errors.Is(err, http.ErrServerClosed) ||
+		strings.Contains(err.Error(), "use of closed") ||
+		strings.Contains(err.Error(), "invalid state") {
+		return nil
+	}
+	return err
+}
+
+// RegisterHandlers registers the /rdp, /ssh, and /vnc routes on mux.
+func (g *Gateway) RegisterHandlers(mux *http.ServeMux) {
+	mux.HandleFunc("/gateway/rdp", g.HandleRDP)
+	mux.HandleFunc("/gateway/ssh", g.HandleSSH)
+	mux.HandleFunc("/gateway/vnc", g.handleVNC)
+}

browsergateway/rdcleanpath.go

@@ -0,0 +1,88 @@
+package browsergateway
+
+import (
+	"encoding/asn1"
+	"fmt"
+)
+
+// RDCleanPath PDU version (BASE_VERSION + 1 = 3389 + 1).
+const rdCleanPathVersion = int64(3390)
+
+// rdCleanPathPdu is a Go translation of the ASN.1 SEQUENCE defined in the
+// ironrdp-rdcleanpath crate. All optional fields use EXPLICIT context-specific
+// tagging, matching the Rust `der::Sequence` derivation with
+// `tag_mode = "EXPLICIT"`.
+//
+// We only need a subset of fields for the basic proxy flow, but the struct
+// declares every tag we may encounter so that decoding does not fail on an
+// unexpected element.
+type rdCleanPathPdu struct {
+	Version           int64    `asn1:"explicit,tag:0"`
+	Destination       string   `asn1:"explicit,tag:2,optional,utf8"`
+	ProxyAuth         string   `asn1:"explicit,tag:3,optional,utf8"`
+	ServerAuth        string   `asn1:"explicit,tag:4,optional,utf8"`
+	PreconnectionBlob string   `asn1:"explicit,tag:5,optional,utf8"`
+	X224              []byte   `asn1:"explicit,tag:6,optional"`
+	ServerCertChain   [][]byte `asn1:"explicit,tag:7,optional"`
+	ServerAddr        string   `asn1:"explicit,tag:9,optional,utf8"`
+}
+
+// decodeRDCleanPathRequest parses a client-to-proxy RDCleanPath PDU.
+func decodeRDCleanPathRequest(buf []byte) (*rdCleanPathPdu, error) {
+	var pdu rdCleanPathPdu
+	rest, err := asn1.Unmarshal(buf, &pdu)
+	if err != nil {
+		return nil, fmt.Errorf("asn1 unmarshal: %w", err)
+	}
+	if len(rest) != 0 {
+		return nil, fmt.Errorf("trailing data after RDCleanPath PDU: %d bytes", len(rest))
+	}
+	if pdu.Version != rdCleanPathVersion {
+		return nil, fmt.Errorf("unexpected RDCleanPath version: %d", pdu.Version)
+	}
+	return &pdu, nil
+}
+
+// encodeRDCleanPathResponse builds a proxy-to-client RDCleanPath response PDU
+// containing the server address, X.224 connection confirm and server TLS chain.
+func encodeRDCleanPathResponse(serverAddr string, x224Rsp []byte, certChain [][]byte) ([]byte, error) {
+	pdu := rdCleanPathPdu{
+		Version:         rdCleanPathVersion,
+		X224:            x224Rsp,
+		ServerCertChain: certChain,
+		ServerAddr:      serverAddr,
+	}
+	return asn1.Marshal(pdu)
+}
+
+// detectRDCleanPathLength returns the total DER length of an RDCleanPath PDU
+// if enough bytes are available, otherwise -1.
+//
+// The PDU is a DER SEQUENCE, which begins with the universal SEQUENCE tag
+// (0x30) followed by a length octet/octets. We parse just enough to know the
+// total length so we can buffer accordingly.
+func detectRDCleanPathLength(buf []byte) int {
+	if len(buf) < 2 {
+		return -1
+	}
+	if buf[0] != 0x30 {
+		// Not a SEQUENCE: cannot be RDCleanPath.
+		return -2
+	}
+	l := buf[1]
+	if l < 0x80 {
+		return 2 + int(l)
+	}
+	n := int(l & 0x7f)
+	if n == 0 || n > 4 {
+		return -2
+	}
+	if len(buf) < 2+n {
+		return -1
+	}
+	total := 0
+	for i := 0; i < n; i++ {
+		total = (total << 8) | int(buf[2+i])
+	}
+	return 2 + n + total
+}

browsergateway/rdp.go

@@ -0,0 +1,271 @@
+package browsergateway
+
+import (
+	"context"
+	"crypto/tls"
+	"encoding/binary"
+	"errors"
+	"fmt"
+	"io"
+	"net"
+	"net/http"
+	"strconv"
+	"time"
+
+	"github.com/coder/websocket"
+	"github.com/fosrl/newt/logger"
+)
+
+// HandleRDP is an http.HandlerFunc for RDP-over-WebSocket connections.
+func (g *Gateway) HandleRDP(w http.ResponseWriter, r *http.Request) {
+	ctx := r.Context()
+	ws, err := websocket.Accept(w, r, &websocket.AcceptOptions{
+		InsecureSkipVerify: true, // any-origin: minimal dev proxy with no auth
+		Subprotocols:       []string{"binary"},
+	})
+	if err != nil {
+		logger.Debug("websocket upgrade failed: %v", err)
+		return
+	}
+	// Disable per-message read size cap (default is 32 KiB which would break
+	// large RDP graphics messages).
+	ws.SetReadLimit(-1)
+	defer ws.CloseNow() //nolint:errcheck
+
+	if err := g.serveSession(ctx, ws); err != nil {
+		logger.Debug("session error: %v", err)
+	}
+}
+
+func (g *Gateway) serveSession(ctx context.Context, ws *websocket.Conn) error {
+	// Expose the WebSocket as a streaming net.Conn. Binary messages are
+	// concatenated into a byte stream and writes become single binary frames.
+	// This is a thin wrapper with no per-message goroutine, unlike Gorilla.
+	stream := websocket.NetConn(ctx, ws, websocket.MessageBinary)
+	defer stream.Close() //nolint:errcheck
+
+	// -- Read the initial RDCleanPath request from the client --
+	pdu, err := readCleanPath(stream)
+	if err != nil {
+		return fmt.Errorf("read RDCleanPath: %w", err)
+	}
+
+	if pdu.Destination == "" {
+		return errors.New("RDCleanPath missing destination")
+	}
+	if len(pdu.X224) == 0 {
+		return errors.New("RDCleanPath missing X224 connection PDU")
+	}
+
+	target := pdu.Destination
+	// Default port for RDP if not specified.
+	if _, _, splitErr := net.SplitHostPort(target); splitErr != nil {
+		target = net.JoinHostPort(target, "3389")
+	}
+
+	// Validate destination against the registered target allowlist,
+	// including per-target auth token.
+	rdpHost, rdpPortStr, _ := net.SplitHostPort(target)
+	rdpPort, _ := strconv.Atoi(rdpPortStr)
+	if !g.isAllowed("rdp", rdpHost, rdpPort, pdu.ProxyAuth) {
+		return fmt.Errorf("RDP destination %s is not in the allowed target list or auth token mismatch", target)
+	}
+
+	logger.Debug("Connecting to RDP server %s", target)
+
+	// -- Open TCP connection to the destination RDP server --
+	serverTCP, err := net.DialTimeout("tcp", target, 15*time.Second)
+	if err != nil {
+		return fmt.Errorf("dial %s: %w", target, err)
+	}
+	defer serverTCP.Close()
+	if tcp, ok := serverTCP.(*net.TCPConn); ok {
+		// NoDelay is Go's default; set explicitly. RDP wants low latency for
+		// input echo, and the bulk path is naturally chunked by TLS records.
+		_ = tcp.SetNoDelay(true)
+		_ = tcp.SetKeepAlive(true)
+		_ = tcp.SetKeepAlivePeriod(30 * time.Second)
+		_ = tcp.SetReadBuffer(forwardBufSize)
+		_ = tcp.SetWriteBuffer(forwardBufSize)
+	}
+	serverAddr := serverTCP.RemoteAddr().String()
+
+	// Forward the optional pre-connection blob, then the X.224 connection request.
+	if pdu.PreconnectionBlob != "" {
+		if _, err := serverTCP.Write([]byte(pdu.PreconnectionBlob)); err != nil {
+			return fmt.Errorf("send PCB: %w", err)
+		}
+	}
+	if _, err := serverTCP.Write(pdu.X224); err != nil {
+		return fmt.Errorf("send X224: %w", err)
+	}
+
+	// -- Read the X.224 connection confirm from the server --
+	x224Rsp, err := readX224(serverTCP)
+	if err != nil {
+		return fmt.Errorf("read X224 response: %w", err)
+	}
+	logX224Negotiation(x224Rsp)
+
+	// -- Upgrade the server connection to TLS (skip verification) --
+	//
+	// Windows RDP hosts are picky: only set SNI when the target is a hostname
+	// (Go would skip SNI for IP literals anyway, but be explicit), and accept
+	// the full range of TLS versions / ciphers since some servers only
+	// negotiate TLS 1.0 or legacy suites.
+	host, _, _ := net.SplitHostPort(target)
+	tlsCfg := &tls.Config{
+		InsecureSkipVerify: true, //nolint:gosec // proxy intentionally skips verification
+		MinVersion:         tls.VersionTLS10,
+		// Cap at TLS 1.2: Windows RDP servers commonly send a TLS "internal_error"
+		// alert when CredSSP/NLA is layered on top of a TLS 1.3 session.
+		MaxVersion: tls.VersionTLS12,
+	}
+	if net.ParseIP(host) == nil {
+		tlsCfg.ServerName = host
+	}
+	tlsConn := tls.Client(serverTCP, tlsCfg)
+	if err := tlsConn.Handshake(); err != nil {
+		return fmt.Errorf("TLS handshake with server: %w", err)
+	}
+	logger.Debug("Server TLS handshake OK (version=0x%04x cipher=0x%04x)",
+		tlsConn.ConnectionState().Version, tlsConn.ConnectionState().CipherSuite)
+
+	// Collect the raw DER server certificate chain to return to the client.
+	state := tlsConn.ConnectionState()
+	if len(state.PeerCertificates) == 0 {
+		return errors.New("server did not present any certificates")
+	}
+	chain := make([][]byte, 0, len(state.PeerCertificates))
+	for _, c := range state.PeerCertificates {
+		chain = append(chain, c.Raw)
+	}
+
+	// -- Send the RDCleanPath response back to the client --
+	rsp, err := encodeRDCleanPathResponse(serverAddr, x224Rsp, chain)
+	if err != nil {
+		return fmt.Errorf("encode RDCleanPath response: %w", err)
+	}
+	if _, err := stream.Write(rsp); err != nil {
+		return fmt.Errorf("write RDCleanPath response: %w", err)
+	}
+
+	logger.Debug("RDCleanPath handshake complete, forwarding traffic to %s", serverAddr)
+
+	// -- Two-way blind forwarding of the (now TLS-encrypted) RDP stream --
+	return forward(stream, tlsConn)
+}
+
+// readCleanPath buffers bytes from the stream until a full RDCleanPath PDU has
+// been received, then decodes it.
+func readCleanPath(r io.Reader) (*rdCleanPathPdu, error) {
+	buf := make([]byte, 0, 1024)
+	tmp := make([]byte, 1024)
+	for {
+		total := detectRDCleanPathLength(buf)
+		switch {
+		case total == -2:
+			return nil, errors.New("invalid RDCleanPath PDU")
+		case total > 0 && len(buf) >= total:
+			return decodeRDCleanPathRequest(buf[:total])
+		}
+		n, err := r.Read(tmp)
+		if n > 0 {
+			buf = append(buf, tmp[:n]...)
+		}
+		if err != nil {
+			return nil, err
+		}
+	}
+}
+
+// readX224 reads exactly one TPKT-framed X.224 PDU from the server.
+//
+// The TPKT header is 4 bytes: version (0x03), reserved (0x00), and a u16
+// big-endian total length that includes the header itself.
+func readX224(r io.Reader) ([]byte, error) {
+	hdr := make([]byte, 4)
+	if _, err := io.ReadFull(r, hdr); err != nil {
+		return nil, err
+	}
+	if hdr[0] != 0x03 {
+		return nil, fmt.Errorf("unexpected TPKT version 0x%02x", hdr[0])
+	}
+	total := int(binary.BigEndian.Uint16(hdr[2:4]))
+	if total < 4 || total > 4096 {
+		return nil, fmt.Errorf("unreasonable TPKT length %d", total)
+	}
+	out := make([]byte, total)
+	copy(out, hdr)
+	if _, err := io.ReadFull(r, out[4:]); err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+// logX224Negotiation prints which RDP security protocol the server selected
+// (or the failure code), to help diagnose handshake issues such as the server
+// requiring NLA/CredSSP.
+//
+// X.224 Connection Confirm layout (RFC 1006 / [MS-RDPBCGR]):
+//
+//	bytes 0..3   TPKT header (03 00 LL LL)
+//	byte  4      X.224 length indicator
+//	byte  5      X.224 code (0xD0 = CC)
+//	bytes 6..10  DST-REF, SRC-REF, class
+//	byte  11     optional RDP Negotiation type (0x02 = response, 0x03 = failure)
+//	byte  12     flags
+//	bytes 13..14 length (little-endian, =8)
+//	bytes 15..18 selected protocol / failure code (u32 little-endian)
+func logX224Negotiation(pdu []byte) {
+	if len(pdu) < 19 {
+		logger.Debug("X.224 response too short (%d bytes) to contain RDP negotiation", len(pdu))
+		return
+	}
+	switch pdu[11] {
+	case 0x02:
+		proto := uint32(pdu[15]) | uint32(pdu[16])<<8 | uint32(pdu[17])<<16 | uint32(pdu[18])<<24
+		name := "unknown"
+		switch proto {
+		case 0:
+			name = "RDP (standard)"
+		case 1:
+			name = "SSL/TLS"
+		case 2:
+			name = "HYBRID (CredSSP/NLA)"
+		case 8:
+			name = "HYBRID_EX"
+		}
+		logger.Debug("Server selected RDP protocol 0x%x (%s)", proto, name)
+	case 0x03:
+		code := uint32(pdu[15]) | uint32(pdu[16])<<8 | uint32(pdu[17])<<16 | uint32(pdu[18])<<24
+		logger.Debug("Server returned RDP negotiation failure code 0x%x", code)
+	default:
+		logger.Debug("X.224 response has no RDP negotiation block (type=0x%02x)", pdu[11])
+	}
+}
+
+// forward shuttles bytes between the two streams until either side closes.
+func forward(a, b io.ReadWriteCloser) error {
+	errc := make(chan error, 2)
+	go func() {
+		buf := make([]byte, forwardBufSize)
+		_, err := io.CopyBuffer(a, b, buf)
+		_ = a.Close()
+		_ = b.Close()
+		errc <- err
+	}()
+	go func() {
+		buf := make([]byte, forwardBufSize)
+		_, err := io.CopyBuffer(b, a, buf)
+		_ = a.Close()
+		_ = b.Close()
+		errc <- err
+	}()
+	// Wait for one side to finish, then return.
+	err := <-errc
+	if errors.Is(err, io.EOF) || err == nil {
+		return nil
+	}
+	return err
+}

browsergateway/ssh.go

@@ -0,0 +1,282 @@
+package browsergateway
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"net"
+	"net/http"
+	"strconv"
+	"time"
+
+	"github.com/coder/websocket"
+	"github.com/fosrl/newt/logger"
+	"golang.org/x/crypto/ssh"
+)
+
+// sshClientMsg is a JSON message sent from the browser to the proxy.
+type sshClientMsg struct {
+	// type: "auth" | "data" | "resize"
+	Type        string `json:"type"`
+	Password    string `json:"password,omitempty"`    // used when type="auth"
+	PrivateKey  string `json:"privateKey,omitempty"`  // used when type="auth"
+	Certificate string `json:"certificate,omitempty"` // used when type="auth"
+	Data        string `json:"data,omitempty"`        // used when type="data"
+	Cols        uint32 `json:"cols,omitempty"`        // used when type="resize"
+	Rows        uint32 `json:"rows,omitempty"`        // used when type="resize"
+}
+
+// sshServerMsg is a JSON message sent from the proxy back to the browser.
+type sshServerMsg struct {
+	// type: "data" | "error"
+	Type  string `json:"type"`
+	Data  string `json:"data,omitempty"`
+	Error string `json:"error,omitempty"`
+}
+
+// HandleSSH is an http.HandlerFunc for SSH-over-WebSocket connections.
+func (g *Gateway) HandleSSH(w http.ResponseWriter, r *http.Request) {
+	logger.Debug("SSH connection request from %s", r.RemoteAddr)
+	ctx := r.Context()
+
+	token := r.URL.Query().Get("authToken")
+
+	// "mode=native" (default) connects to the local SSH daemon on this host.
+	// "mode=proxy" connects to an arbitrary host+port supplied in query params.
+	nativeSSH := r.URL.Query().Get("mode") != "proxy"
+
+	// In proxy mode we also need host + username from query params.
+	var target, username string
+	if !nativeSSH {
+		host := r.URL.Query().Get("host")
+		port := r.URL.Query().Get("port")
+		username = r.URL.Query().Get("username")
+		if host == "" || username == "" {
+			http.Error(w, "missing host or username", http.StatusBadRequest)
+			return
+		}
+		if port == "" {
+			port = "22"
+		}
+		sshPort, _ := strconv.Atoi(port)
+		if !g.isAllowed("ssh", host, sshPort, token) {
+			http.Error(w, "destination not allowed or auth token mismatch", http.StatusForbidden)
+			return
+		}
+		target = net.JoinHostPort(host, port)
+	} else {
+		// Native SSH mode: validate the token against any registered ssh target.
+		if !g.isTokenValid("ssh", token) {
+			http.Error(w, "unauthorized", http.StatusUnauthorized)
+			return
+		}
+		username = r.URL.Query().Get("username")
+		if username == "" {
+			http.Error(w, "missing username", http.StatusBadRequest)
+			return
+		}
+	}
+
+	ws, err := websocket.Accept(w, r, &websocket.AcceptOptions{
+		InsecureSkipVerify: true,
+		Subprotocols:       []string{"ssh"},
+	})
+	if err != nil {
+		logger.Debug("SSH websocket upgrade failed: %v", err)
+		return
+	}
+	ws.SetReadLimit(-1)
+	defer ws.CloseNow() //nolint:errcheck
+
+	if nativeSSH {
+		if err := serveNativeSSHSession(ctx, ws, username, g.sshCreds); err != nil {
+			logger.Debug("SSH native session error: %v", err)
+		}
+	} else {
+		if err := serveSSHSession(ctx, ws, target, username, g.authToken); err != nil {
+			logger.Debug("SSH session error: %v", err)
+		}
+	}
+}
+
+func serveSSHSession(ctx context.Context, ws *websocket.Conn, target, username, _ string) error {
+	// -- Wait for the auth message from the client to get the password --
+	_, authBytes, err := ws.Read(ctx)
+	if err != nil {
+		return fmt.Errorf("read auth message: %w", err)
+	}
+	var authMsg sshClientMsg
+	if err := json.Unmarshal(authBytes, &authMsg); err != nil || authMsg.Type != "auth" {
+		return fmt.Errorf("expected auth message, got: %s", authBytes)
+	}
+	password := authMsg.Password
+	privateKey := authMsg.PrivateKey
+	certificate := authMsg.Certificate
+
+	// Build the list of auth methods. Certificate+private key takes priority
+	// when both are provided, then private key, then password.
+	var authMethods []ssh.AuthMethod
+	if privateKey != "" {
+		signer, err := ssh.ParsePrivateKey([]byte(privateKey))
+		if err != nil {
+			sendSSHError(ctx, ws, fmt.Sprintf("Failed to parse private key: %v", err))
+			return fmt.Errorf("parse private key: %w", err)
+		}
+
+		if certificate != "" {
+			certPubKey, _, _, _, err := ssh.ParseAuthorizedKey([]byte(certificate))
+			if err != nil {
+				sendSSHError(ctx, ws, fmt.Sprintf("Failed to parse certificate: %v", err))
+				return fmt.Errorf("parse certificate: %w", err)
+			}
+
+			cert, ok := certPubKey.(*ssh.Certificate)
+			if !ok {
+				sendSSHError(ctx, ws, "Provided certificate is not a valid SSH certificate")
+				return fmt.Errorf("provided certificate is not ssh certificate")
+			}
+
+			certSigner, err := ssh.NewCertSigner(cert, signer)
+			if err != nil {
+				sendSSHError(ctx, ws, fmt.Sprintf("Failed to combine private key and certificate: %v", err))
+				return fmt.Errorf("new cert signer: %w", err)
+			}
+			authMethods = append(authMethods, ssh.PublicKeys(certSigner))
+		} else {
+			authMethods = append(authMethods, ssh.PublicKeys(signer))
+		}
+	}
+	if password != "" {
+		authMethods = append(authMethods, ssh.Password(password))
+	}
+	if len(authMethods) == 0 {
+		sendSSHError(ctx, ws, "No authentication credentials provided")
+		return fmt.Errorf("no auth credentials")
+	}
+	logger.Debug("SSH: connecting to %s as %s", target, username)
+	sshCfg := &ssh.ClientConfig{
+		User: username,
+		Auth: authMethods,
+		// HostKeyCallback is intentionally InsecureIgnoreHostKey for this dev
+		// proxy. In production, verify against a known-hosts store.
+		HostKeyCallback: ssh.InsecureIgnoreHostKey(), //nolint:gosec
+		Timeout:         15 * time.Second,
+	}
+
+	sshClient, err := ssh.Dial("tcp", target, sshCfg)
+	if err != nil {
+		sendSSHError(ctx, ws, fmt.Sprintf("SSH dial failed: %v", err))
+		return fmt.Errorf("ssh dial %s: %w", target, err)
+	}
+	defer sshClient.Close()
+
+	// -- Open an interactive session --
+	sess, err := sshClient.NewSession()
+	if err != nil {
+		sendSSHError(ctx, ws, fmt.Sprintf("Failed to open SSH session: %v", err))
+		return fmt.Errorf("ssh new session: %w", err)
+	}
+	defer sess.Close()
+
+	// Request a PTY.
+	if err := sess.RequestPty("xterm-256color", 24, 80, ssh.TerminalModes{
+		ssh.ECHO:          1,
+		ssh.TTY_OP_ISPEED: 38400,
+		ssh.TTY_OP_OSPEED: 38400,
+	}); err != nil {
+		sendSSHError(ctx, ws, fmt.Sprintf("Failed to request PTY: %v", err))
+		return fmt.Errorf("ssh request pty: %w", err)
+	}
+
+	stdinPipe, err := sess.StdinPipe()
+	if err != nil {
+		return fmt.Errorf("ssh stdin pipe: %w", err)
+	}
+	stdoutPipe, err := sess.StdoutPipe()
+	if err != nil {
+		return fmt.Errorf("ssh stdout pipe: %w", err)
+	}
+	stderrPipe, err := sess.StderrPipe()
+	if err != nil {
+		return fmt.Errorf("ssh stderr pipe: %w", err)
+	}
+
+	if err := sess.Shell(); err != nil {
+		sendSSHError(ctx, ws, fmt.Sprintf("Failed to start shell: %v", err))
+		return fmt.Errorf("ssh shell: %w", err)
+	}
+
+	logger.Debug("SSH: session established with %s", target)
+
+	// -- Pump SSH stdout/stderr → WebSocket --
+	sessCtx, cancelSess := context.WithCancel(ctx)
+	defer cancelSess()
+
+	go func() {
+		buf := make([]byte, 4096)
+		for {
+			n, readErr := stdoutPipe.Read(buf)
+			if n > 0 {
+				msg := sshServerMsg{Type: "data", Data: string(buf[:n])}
+				b, _ := json.Marshal(msg)
+				if writeErr := ws.Write(sessCtx, websocket.MessageText, b); writeErr != nil {
+					return
+				}
+			}
+			if readErr != nil {
+				cancelSess()
+				return
+			}
+		}
+	}()
+
+	go func() {
+		buf := make([]byte, 4096)
+		for {
+			n, readErr := stderrPipe.Read(buf)
+			if n > 0 {
+				msg := sshServerMsg{Type: "data", Data: string(buf[:n])}
+				b, _ := json.Marshal(msg)
+				if writeErr := ws.Write(sessCtx, websocket.MessageText, b); writeErr != nil {
+					return
+				}
+			}
+			if readErr != nil {
+				return
+			}
+		}
+	}()
+
+	// -- Pump WebSocket input → SSH stdin / resize --
+	for {
+		_, msgBytes, readErr := ws.Read(sessCtx)
+		if readErr != nil {
+			break
+		}
+
+		var msg sshClientMsg
+		if err := json.Unmarshal(msgBytes, &msg); err != nil {
+			continue
+		}
+
+		switch msg.Type {
+		case "data":
+			if _, err := stdinPipe.Write([]byte(msg.Data)); err != nil {
+				return fmt.Errorf("write ssh stdin: %w", err)
+			}
+		case "resize":
+			if msg.Cols > 0 && msg.Rows > 0 {
+				_ = sess.WindowChange(int(msg.Rows), int(msg.Cols))
+			}
+		}
+	}
+
+	return nil
+}
+
+// sendSSHError sends an error message to the browser and logs it.
+func sendSSHError(ctx context.Context, ws *websocket.Conn, msg string) {
+	logger.Debug("SSH error: %s", msg)
+	b, _ := json.Marshal(sshServerMsg{Type: "error", Error: msg})
+	_ = ws.Write(ctx, websocket.MessageText, b)
+}

browsergateway/ssh_native.go

@@ -0,0 +1,94 @@
+//go:build !windows
+
+package browsergateway
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+
+	"github.com/coder/websocket"
+	"github.com/fosrl/newt/logger"
+	"github.com/fosrl/newt/nativessh"
+)
+
+// serveNativeSSHSession handles a WebSocket SSH session by authenticating the
+// user against the host OS (authorized_keys then PAM password) and then
+// spawning a PTY+shell running as that user.
+//
+// The auth frame from the browser must be a JSON sshClientMsg with type="auth"
+// carrying the same password/privateKey fields used by the proxy SSH path.
+// The target username is passed in from the HTTP layer (query param).
+func serveNativeSSHSession(ctx context.Context, ws *websocket.Conn, username string, creds *nativessh.CredentialStore) error {
+	// Read the auth frame.
+	_, authBytes, err := ws.Read(ctx)
+	if err != nil {
+		return fmt.Errorf("read auth message: %w", err)
+	}
+	var authMsg sshClientMsg
+	if err := json.Unmarshal(authBytes, &authMsg); err != nil || authMsg.Type != "auth" {
+		return fmt.Errorf("expected auth message, got: %s", authBytes)
+	}
+
+	// Authenticate using host authorized_keys or PAM password.
+	if err := nativessh.AuthenticateWithCertificate(creds, username, authMsg.Password, authMsg.PrivateKey, authMsg.Certificate); err != nil {
+		sendSSHError(ctx, ws, "Authentication failed")
+		return fmt.Errorf("auth for user %q: %w", username, err)
+	}
+
+	logger.Debug("SSH native: spawning shell as user %q", username)
+
+	sess, err := nativessh.NewPTYSessionAs(username)
+	if err != nil {
+		sendSSHError(ctx, ws, fmt.Sprintf("Failed to spawn shell: %v", err))
+		return fmt.Errorf("pty session as %q: %w", username, err)
+	}
+	defer sess.Close()
+
+	// Cancel context to unblock the WebSocket read loop when the shell exits.
+	sessCtx, cancelSess := context.WithCancel(ctx)
+	defer cancelSess()
+
+	// Pump PTY output → WebSocket.
+	go func() {
+		defer cancelSess()
+		buf := make([]byte, 4096)
+		for {
+			n, readErr := sess.Read(buf)
+			if n > 0 {
+				msg := sshServerMsg{Type: "data", Data: string(buf[:n])}
+				b, _ := json.Marshal(msg)
+				if writeErr := ws.Write(sessCtx, websocket.MessageText, b); writeErr != nil {
+					return
+				}
+			}
+			if readErr != nil {
+				return
+			}
+		}
+	}()
+
+	// Pump WebSocket input → PTY stdin / resize.
+	for {
+		_, msgBytes, readErr := ws.Read(sessCtx)
+		if readErr != nil {
+			break
+		}
+		var msg sshClientMsg
+		if err := json.Unmarshal(msgBytes, &msg); err != nil {
+			continue
+		}
+		switch msg.Type {
+		case "data":
+			if _, writeErr := sess.Write([]byte(msg.Data)); writeErr != nil {
+				return fmt.Errorf("write pty: %w", writeErr)
+			}
+		case "resize":
+			if msg.Cols > 0 && msg.Rows > 0 {
+				_ = sess.Resize(uint16(msg.Cols), uint16(msg.Rows))
+			}
+		}
+	}
+
+	return nil
+}

browsergateway/ssh_native_windows.go

@@ -0,0 +1,16 @@
+//go:build windows
+
+package browsergateway
+
+import (
+	"context"
+	"errors"
+
+	"github.com/coder/websocket"
+	"github.com/fosrl/newt/nativessh"
+)
+
+// serveNativeSSHSession is not supported on Windows.
+func serveNativeSSHSession(_ context.Context, _ *websocket.Conn, _ string, _ *nativessh.CredentialStore) error {
+	return errors.New("native SSH is not supported on Windows")
+}

browsergateway/vnc.go

@@ -0,0 +1,125 @@
+package browsergateway
+
+import (
+	"context"
+	"fmt"
+	"io"
+	"net"
+	"net/http"
+	"strconv"
+	"time"
+
+	"github.com/coder/websocket"
+	"github.com/fosrl/newt/logger"
+)
+
+const (
+	vncDialTimeout    = 10 * time.Second
+	vncKeepAlive      = 30 * time.Second
+	vncForwardBufSize = 32 * 1024
+)
+
+// handleVNC proxies a noVNC WebSocket connection to a raw TCP VNC backend.
+// It follows the same auth-token-in-query-param pattern as handleSSH.
+//
+// Query parameters:
+//
+//	authToken – shared secret matching the -auth-token flag
+//	host      – VNC backend hostname or IP
+//	port      – VNC backend port (default: 5900)
+func (g *Gateway) handleVNC(w http.ResponseWriter, r *http.Request) {
+	host := r.URL.Query().Get("host")
+	port := r.URL.Query().Get("port")
+	if host == "" {
+		http.Error(w, "missing host", http.StatusBadRequest)
+		return
+	}
+	if port == "" {
+		port = "5900"
+	}
+	vncPort, _ := strconv.Atoi(port)
+	authToken := r.URL.Query().Get("authToken")
+	if !g.isAllowed("vnc", host, vncPort, authToken) {
+		http.Error(w, "destination not allowed or auth token mismatch", http.StatusForbidden)
+		return
+	}
+	target := net.JoinHostPort(host, port)
+
+	// Optional HTTP probe used by the web UI to surface backend reachability
+	// failures before attempting a WebSocket session.
+	if r.URL.Query().Get("checkOnly") == "1" {
+		if err := dialVNCBackend(r.Context(), target); err != nil {
+			logger.Debug("vnc: preflight failed (%s): %v", target, err)
+			http.Error(w, fmt.Sprintf("failed to connect to VNC backend: %v", err), http.StatusBadGateway)
+			return
+		}
+		w.WriteHeader(http.StatusNoContent)
+		return
+	}
+
+	// Accept the WebSocket. noVNC negotiates the "binary" subprotocol;
+	// fall back gracefully when the client sends "base64" as well.
+	ws, err := websocket.Accept(w, r, &websocket.AcceptOptions{
+		InsecureSkipVerify: true,
+		Subprotocols:       []string{"binary", "base64"},
+	})
+	if err != nil {
+		logger.Debug("vnc: websocket upgrade failed: %v", err)
+		return
+	}
+	ws.SetReadLimit(-1)
+	defer ws.CloseNow() //nolint:errcheck
+
+	ctx := r.Context()
+	if err := serveVNC(ctx, ws, target); err != nil {
+		logger.Debug("vnc: session error (%s): %v", target, err)
+	}
+}
+
+func dialVNCBackend(ctx context.Context, target string) error {
+	dialer := &net.Dialer{
+		Timeout:   vncDialTimeout,
+		KeepAlive: vncKeepAlive,
+	}
+	conn, err := dialer.DialContext(ctx, "tcp", target)
+	if err != nil {
+		return err
+	}
+	defer conn.Close() //nolint:errcheck
+
+	return nil
+}
+
+func serveVNC(ctx context.Context, ws *websocket.Conn, target string) error {
+	// Dial the VNC backend TCP server.
+	dialer := &net.Dialer{
+		Timeout:   vncDialTimeout,
+		KeepAlive: vncKeepAlive,
+	}
+	conn, err := dialer.DialContext(ctx, "tcp", target)
+	if err != nil {
+		return err
+	}
+	defer conn.Close() //nolint:errcheck
+
+	// Expose the WebSocket as a plain net.Conn byte stream (binary frames).
+	stream := websocket.NetConn(ctx, ws, websocket.MessageBinary)
+	defer stream.Close() //nolint:errcheck
+
+	// Proxy bidirectionally: VNC backend <-> browser.
+	errc := make(chan error, 2)
+	go func() {
+		buf := make([]byte, vncForwardBufSize)
+		_, err := io.CopyBuffer(conn, stream, buf)
+		errc <- err
+	}()
+	go func() {
+		buf := make([]byte, vncForwardBufSize)
+		_, err := io.CopyBuffer(stream, conn, buf)
+		errc <- err
+	}()
+
+	// Return when either direction closes.
+	err = <-errc
+	return err
+}

clients.go

@@ -0,0 +1,114 @@
+package main
+
+import (
+	"strings"
+
+	"github.com/fosrl/newt/clients"
+	wgnetstack "github.com/fosrl/newt/clients"
+	"github.com/fosrl/newt/clients/permissions"
+	"github.com/fosrl/newt/logger"
+	"github.com/fosrl/newt/nativessh"
+	"github.com/fosrl/newt/websocket"
+	"golang.zx2c4.com/wireguard/tun/netstack"
+)
+
+var wgService *clients.WireGuardService
+var ready bool
+
+func setupClients(client *websocket.Client, credStore *nativessh.CredentialStore) {
+	var host = endpoint
+	if strings.HasPrefix(host, "http://") {
+		host = strings.TrimPrefix(host, "http://")
+	} else if strings.HasPrefix(host, "https://") {
+		host = strings.TrimPrefix(host, "https://")
+	}
+
+	host = strings.TrimSuffix(host, "/")
+
+	logger.Debug("Setting up clients with netstack2...")
+
+	// if useNativeInterface is true make sure we have permission to use native interface
+	if useNativeInterface {
+		logger.Debug("Checking permissions for native interface")
+		err := permissions.CheckNativeInterfacePermissions()
+		if err != nil {
+			logger.Fatal("Insufficient permissions to create native TUN interface: %v", err)
+			return
+		}
+	}
+
+	// Create WireGuard service
+	wgService, err = wgnetstack.NewWireGuardService(interfaceName, port, mtuInt, host, id, client, dns, useNativeInterface)
+	if err != nil {
+		logger.Fatal("Failed to create WireGuard service: %v", err)
+	}
+
+	wgService.SetCredentialStore(credStore)
+
+	client.OnTokenUpdate(func(token string) {
+		wgService.SetToken(token)
+	})
+
+	ready = true
+}
+
+func setDownstreamTNetstack(tnet *netstack.Net) {
+	if wgService != nil {
+		wgService.SetOthertnet(tnet)
+	}
+}
+
+func closeClients() {
+	logger.Info("Closing clients...")
+	if wgService != nil {
+		wgService.Close()
+		wgService = nil
+	}
+}
+
+// setClientsBlocked enables or disables connection blocking on the WireGuard service.
+func setClientsBlocked(v bool) {
+	if wgService != nil {
+		wgService.SetBlocked(v)
+	}
+}
+
+func clientsHandleNewtConnection(publicKey string, endpoint string, relayPort uint16) {
+	if !ready {
+		return
+	}
+
+	// split off the port from the endpoint
+	parts := strings.Split(endpoint, ":")
+	if len(parts) < 2 {
+		logger.Error("Invalid endpoint format: %s", endpoint)
+		return
+	}
+	endpoint = strings.Join(parts[:len(parts)-1], ":")
+
+	if wgService != nil {
+		wgService.StartHolepunch(publicKey, endpoint, relayPort)
+	}
+}
+
+func clientsOnConnect() {
+	if !ready {
+		return
+	}
+	if wgService != nil {
+		wgService.LoadRemoteConfig()
+	}
+}
+
+// clientsStartDirectRelay starts a direct UDP relay from the main tunnel netstack
+// to the clients' WireGuard, bypassing the proxy for better performance.
+func clientsStartDirectRelay(tunnelIP string) {
+	if !ready {
+		return
+	}
+	if wgService != nil {
+		if err := wgService.StartDirectUDPRelay(tunnelIP); err != nil {
+			logger.Error("Failed to start direct UDP relay: %v", err)
+		}
+	}
+}

clients/permissions/permissions_android.go

@@ -0,0 +1,8 @@
+//go:build android
+
+package permissions
+
+// CheckNativeInterfacePermissions always allows permission on Android.
+func CheckNativeInterfacePermissions() error {
+	return nil
+}

clients/permissions/permissions_darwin.go

@@ -0,0 +1,18 @@
+//go:build darwin && !ios
+
+package permissions
+
+import (
+	"fmt"
+	"os"
+)
+
+// CheckNativeInterfacePermissions checks if the process has sufficient
+// permissions to create a native TUN interface on macOS.
+// This typically requires root privileges.
+func CheckNativeInterfacePermissions() error {
+	if os.Geteuid() == 0 {
+		return nil
+	}
+	return fmt.Errorf("insufficient permissions: need root to create TUN interface on macOS")
+}

clients/permissions/permissions_freebsd.go

@@ -0,0 +1,57 @@
+//go:build freebsd
+
+package permissions
+
+import (
+	"fmt"
+	"os"
+
+	"github.com/fosrl/newt/logger"
+)
+
+const (
+	// TUN device on FreeBSD
+	tunDevice = "/dev/tun"
+	ifnamsiz  = 16
+	iffTun    = 0x0001
+	iffNoPi   = 0x1000
+)
+
+// ifReq is the structure for TUN interface configuration
+type ifReq struct {
+	Name  [ifnamsiz]byte
+	Flags uint16
+	_     [22]byte // padding to match kernel structure
+}
+
+// CheckNativeInterfacePermissions checks if the process has sufficient
+// permissions to create a native TUN interface on FreeBSD.
+// This requires root privileges (UID 0).
+func CheckNativeInterfacePermissions() error {
+	logger.Debug("Checking native interface permissions on FreeBSD")
+
+	// Check if running as root
+	if os.Geteuid() == 0 {
+		logger.Debug("Running as root, sufficient permissions for native TUN interface")
+		return nil
+	}
+
+	// On FreeBSD, only root can create TUN interfaces
+	// Try to open the TUN device to verify
+	return tryOpenTunDevice()
+}
+
+// tryOpenTunDevice attempts to open the TUN device to verify permissions.
+// On FreeBSD, /dev/tun is a cloning device that creates a new interface
+// when opened.
+func tryOpenTunDevice() error {
+	// Try opening /dev/tun (cloning device)
+	f, err := os.OpenFile(tunDevice, os.O_RDWR, 0)
+	if err != nil {
+		return fmt.Errorf("cannot open %s: %v (need root privileges)", tunDevice, err)
+	}
+	defer f.Close()
+
+	logger.Debug("Successfully opened TUN device, sufficient permissions for native TUN interface")
+	return nil
+}

clients/permissions/permissions_ios.go

@@ -0,0 +1,8 @@
+//go:build ios
+
+package permissions
+
+// CheckNativeInterfacePermissions always allows permission on iOS.
+func CheckNativeInterfacePermissions() error {
+	return nil
+}

clients/permissions/permissions_linux.go

@@ -0,0 +1,96 @@
+//go:build linux && !android
+
+package permissions
+
+import (
+	"fmt"
+	"os"
+	"unsafe"
+
+	"github.com/fosrl/newt/logger"
+	"golang.org/x/sys/unix"
+)
+
+const (
+	// TUN device constants
+	tunDevice = "/dev/net/tun"
+	ifnamsiz  = 16
+	iffTun    = 0x0001
+	iffNoPi   = 0x1000
+	tunSetIff = 0x400454ca
+)
+
+// ifReq is the structure for TUNSETIFF ioctl
+type ifReq struct {
+	Name  [ifnamsiz]byte
+	Flags uint16
+	_     [22]byte // padding to match kernel structure
+}
+
+// CheckNativeInterfacePermissions checks if the process has sufficient
+// permissions to create a native TUN interface on Linux.
+// This requires either root privileges (UID 0) or CAP_NET_ADMIN capability.
+func CheckNativeInterfacePermissions() error {
+	logger.Debug("Checking native interface permissions on Linux")
+
+	// Check if running as root
+	if os.Geteuid() == 0 {
+		logger.Debug("Running as root, sufficient permissions for native TUN interface")
+		return nil
+	}
+
+	// Check for CAP_NET_ADMIN capability
+	caps := unix.CapUserHeader{
+		Version: unix.LINUX_CAPABILITY_VERSION_3,
+		Pid:     0, // 0 means current process
+	}
+
+	var data [2]unix.CapUserData
+	if err := unix.Capget(&caps, &data[0]); err != nil {
+		logger.Debug("Failed to get capabilities: %v, will try creating test TUN", err)
+	} else {
+		// CAP_NET_ADMIN is capability bit 12
+		const CAP_NET_ADMIN = 12
+		if data[0].Effective&(1<<CAP_NET_ADMIN) != 0 {
+			logger.Debug("Process has CAP_NET_ADMIN capability, sufficient permissions for native TUN interface")
+			return nil
+		}
+		logger.Debug("Process does not have CAP_NET_ADMIN capability in effective set")
+	}
+
+	// Actually try to create a TUN interface to verify permissions
+	// This is the most reliable check as it tests the actual operation
+	return tryCreateTestTun()
+}
+
+// tryCreateTestTun attempts to create a temporary TUN interface to verify
+// we have the necessary permissions. This tests the actual ioctl call that
+// will be used when creating the real interface.
+func tryCreateTestTun() error {
+	f, err := os.OpenFile(tunDevice, os.O_RDWR, 0)
+	if err != nil {
+		return fmt.Errorf("cannot open %s: %v (need root or CAP_NET_ADMIN capability)", tunDevice, err)
+	}
+	defer f.Close()
+
+	// Try to create a TUN interface with a test name
+	// Using a random-ish name to avoid conflicts
+	var req ifReq
+	copy(req.Name[:], "tuntest0")
+	req.Flags = iffTun | iffNoPi
+
+	_, _, errno := unix.Syscall(
+		unix.SYS_IOCTL,
+		f.Fd(),
+		uintptr(tunSetIff),
+		uintptr(unsafe.Pointer(&req)),
+	)
+
+	if errno != 0 {
+		return fmt.Errorf("cannot create TUN interface (ioctl TUNSETIFF failed): %v (need root or CAP_NET_ADMIN capability)", errno)
+	}
+
+	// Success - the interface will be automatically destroyed when we close the fd
+	logger.Debug("Successfully created test TUN interface, sufficient permissions for native TUN interface")
+	return nil
+}

clients/permissions/permissions_windows.go

@@ -0,0 +1,48 @@
+//go:build windows
+
+package permissions
+
+import (
+	"fmt"
+
+	"golang.org/x/sys/windows"
+	"golang.org/x/sys/windows/svc"
+)
+
+// CheckNativeInterfacePermissions checks if the process has sufficient
+// permissions to create a native TUN interface on Windows.
+// This requires Administrator privileges and must be running as a Windows service.
+func CheckNativeInterfacePermissions() error {
+	// Check if running as a Windows service
+	isService, err := svc.IsWindowsService()
+	if err != nil {
+		return fmt.Errorf("failed to check if running as Windows service: %v", err)
+	}
+	if !isService {
+		return fmt.Errorf("native TUN interface requires running as a Windows service")
+	}
+
+	var sid *windows.SID
+	err = windows.AllocateAndInitializeSid(
+		&windows.SECURITY_NT_AUTHORITY,
+		2,
+		windows.SECURITY_BUILTIN_DOMAIN_RID,
+		windows.DOMAIN_ALIAS_RID_ADMINS,
+		0, 0, 0, 0, 0, 0,
+		&sid)
+	if err != nil {
+		return fmt.Errorf("failed to initialize SID: %v", err)
+	}
+	defer windows.FreeSid(sid)
+
+	token := windows.Token(0)
+	member, err := token.IsMember(sid)
+	if err != nil {
+		return fmt.Errorf("failed to check admin group membership: %v", err)
+	}
+
+	if !member {
+		return fmt.Errorf("insufficient permissions: need Administrator to create TUN interface on Windows")
+	}
+	return nil
+}

common.go

@@ -0,0 +1,661 @@
+package main
+
+import (
+	"bytes"
+	"context"
+	"encoding/json"
+	"fmt"
+	"net"
+	"os"
+	"os/exec"
+	"regexp"
+	"strings"
+	"time"
+
+	"math/rand"
+
+	"github.com/fosrl/newt/internal/telemetry"
+	"github.com/fosrl/newt/logger"
+	"github.com/fosrl/newt/proxy"
+	"github.com/fosrl/newt/websocket"
+	"github.com/fsnotify/fsnotify"
+	"golang.org/x/net/icmp"
+	"golang.org/x/net/ipv4"
+	"golang.zx2c4.com/wireguard/tun/netstack"
+	"gopkg.in/yaml.v3"
+)
+
+const msgHealthFileWriteFailed = "Failed to write health file: %v"
+
+func ping(tnet *netstack.Net, dst string, timeout time.Duration) (time.Duration, error) {
+	// logger.Debug("Pinging %s", dst)
+	socket, err := tnet.Dial("ping4", dst)
+	if err != nil {
+		return 0, fmt.Errorf("failed to create ICMP socket: %w", err)
+	}
+	defer socket.Close()
+
+	// Set socket buffer sizes to handle high bandwidth scenarios
+	if tcpConn, ok := socket.(interface{ SetReadBuffer(int) error }); ok {
+		tcpConn.SetReadBuffer(64 * 1024)
+	}
+	if tcpConn, ok := socket.(interface{ SetWriteBuffer(int) error }); ok {
+		tcpConn.SetWriteBuffer(64 * 1024)
+	}
+
+	requestPing := icmp.Echo{
+		Seq:  rand.Intn(1 << 16),
+		Data: []byte("newtping"),
+	}
+
+	icmpBytes, err := (&icmp.Message{Type: ipv4.ICMPTypeEcho, Code: 0, Body: &requestPing}).Marshal(nil)
+	if err != nil {
+		return 0, fmt.Errorf("failed to marshal ICMP message: %w", err)
+	}
+
+	if err := socket.SetReadDeadline(time.Now().Add(timeout)); err != nil {
+		return 0, fmt.Errorf("failed to set read deadline: %w", err)
+	}
+
+	start := time.Now()
+	_, err = socket.Write(icmpBytes)
+	if err != nil {
+		return 0, fmt.Errorf("failed to write ICMP packet: %w", err)
+	}
+
+	// Use larger buffer for reading to handle potential network congestion
+	readBuffer := make([]byte, 1500)
+	n, err := socket.Read(readBuffer)
+	if err != nil {
+		return 0, fmt.Errorf("failed to read ICMP packet: %w", err)
+	}
+
+	replyPacket, err := icmp.ParseMessage(1, readBuffer[:n])
+	if err != nil {
+		return 0, fmt.Errorf("failed to parse ICMP packet: %w", err)
+	}
+
+	replyPing, ok := replyPacket.Body.(*icmp.Echo)
+	if !ok {
+		return 0, fmt.Errorf("invalid reply type: got %T, want *icmp.Echo", replyPacket.Body)
+	}
+
+	if !bytes.Equal(replyPing.Data, requestPing.Data) || replyPing.Seq != requestPing.Seq {
+		return 0, fmt.Errorf("invalid ping reply: got seq=%d data=%q, want seq=%d data=%q",
+			replyPing.Seq, replyPing.Data, requestPing.Seq, requestPing.Data)
+	}
+
+	latency := time.Since(start)
+
+	// logger.Debug("Ping to %s successful, latency: %v", dst, latency)
+
+	return latency, nil
+}
+
+// reliablePing performs multiple ping attempts with adaptive timeout
+func reliablePing(tnet *netstack.Net, dst string, baseTimeout time.Duration, maxAttempts int) (time.Duration, error) {
+	var lastErr error
+	var totalLatency time.Duration
+	successCount := 0
+
+	for attempt := 1; attempt <= maxAttempts; attempt++ {
+		// Adaptive timeout: increase timeout for later attempts
+		timeout := baseTimeout + time.Duration(attempt-1)*500*time.Millisecond
+
+		// Add jitter to prevent thundering herd
+		jitter := time.Duration(rand.Intn(100)) * time.Millisecond
+		timeout += jitter
+
+		latency, err := ping(tnet, dst, timeout)
+		if err != nil {
+			lastErr = err
+			logger.Debug("Ping attempt %d/%d failed: %v", attempt, maxAttempts, err)
+
+			// Brief pause between attempts with exponential backoff
+			if attempt < maxAttempts {
+				backoff := time.Duration(attempt) * 50 * time.Millisecond
+				time.Sleep(backoff)
+			}
+			continue
+		}
+
+		totalLatency += latency
+		successCount++
+
+		// Return on first success
+		return totalLatency / time.Duration(successCount), nil
+	}
+
+	return 0, fmt.Errorf("all %d ping attempts failed, last error: %v", maxAttempts, lastErr)
+}
+
+func pingWithRetry(tnet *netstack.Net, dst string, timeout time.Duration) (stopChan chan struct{}, err error) {
+
+	if healthFile != "" {
+		err = os.Remove(healthFile)
+		if err != nil {
+			logger.Error("Failed to remove health file: %v", err)
+		}
+	}
+
+	const (
+		initialMaxAttempts = 5
+		initialRetryDelay  = 2 * time.Second
+		maxRetryDelay      = 60 * time.Second // Cap the maximum delay
+	)
+
+	stopChan = make(chan struct{})
+	attempt := 1
+	retryDelay := initialRetryDelay
+
+	// First try with the initial parameters
+	logger.Debug("Ping attempt %d", attempt)
+	if latency, err := ping(tnet, dst, timeout); err == nil {
+		// Successful ping
+		logger.Debug("Ping latency: %v", latency)
+		logger.Info("Tunnel connection to server established successfully!")
+		if healthFile != "" {
+			err := os.WriteFile(healthFile, []byte("ok"), 0644)
+			if err != nil {
+				logger.Warn(msgHealthFileWriteFailed, err)
+			}
+		}
+		return stopChan, nil
+	} else {
+		logger.Warn("Ping attempt %d failed: %v", attempt, err)
+	}
+
+	// Start a goroutine that will attempt pings indefinitely with increasing delays
+	go func() {
+		attempt = 2 // Continue from attempt 2
+
+		for {
+			select {
+			case <-stopChan:
+				return
+			default:
+				logger.Debug("Ping attempt %d", attempt)
+
+				if latency, err := ping(tnet, dst, timeout); err != nil {
+					logger.Warn("Ping attempt %d failed: %v", attempt, err)
+
+					// Increase delay after certain thresholds but cap it
+					if attempt%5 == 0 && retryDelay < maxRetryDelay {
+						retryDelay = time.Duration(float64(retryDelay) * 1.5)
+						if retryDelay > maxRetryDelay {
+							retryDelay = maxRetryDelay
+						}
+						logger.Info("Increasing ping retry delay to %v", retryDelay)
+					}
+
+					time.Sleep(retryDelay)
+					attempt++
+				} else {
+					// Successful ping
+					logger.Debug("Ping succeeded after %d attempts", attempt)
+					logger.Debug("Ping latency: %v", latency)
+					logger.Info("Tunnel connection to server established successfully!")
+					if healthFile != "" {
+						err := os.WriteFile(healthFile, []byte("ok"), 0644)
+						if err != nil {
+							logger.Warn(msgHealthFileWriteFailed, err)
+						}
+					}
+					return
+				}
+			case <-pingStopChan:
+				// Stop the goroutine when signaled
+				return
+			}
+		}
+	}()
+
+	// Return an error for the first batch of attempts (to maintain compatibility with existing code)
+	return stopChan, fmt.Errorf("initial ping attempts failed, continuing in background")
+}
+
+// shouldFireRecovery decides whether the data-plane recovery flow in
+// startPingCheck should run on this tick. Recovery fires once when the
+// consecutive-failure counter first crosses the threshold; the connectionLost
+// flag prevents re-firing until a successful ping resets the state.
+//
+// This condition was previously inlined into startPingCheck and AND-ed with
+// `currentInterval < maxInterval`, which silently broke recovery once
+// pingInterval's default was bumped to 15s while maxInterval stayed at 6s
+// (commit 8161fa6, March 2026): the gate became permanently false on default
+// settings, so the recovery code never executed and ping failures climbed
+// forever — the proximate cause of fosrl/newt#284, #310 and pangolin#1004.
+//
+// Recovery and backoff are independent concerns; the backoff ramp is now
+// computed separately in the caller. Do not re-introduce currentInterval
+// here.
+func shouldFireRecovery(consecutiveFailures, failureThreshold int, connectionLost bool) bool {
+	return consecutiveFailures >= failureThreshold && !connectionLost
+}
+
+func startPingCheck(tnet *netstack.Net, serverIP string, client *websocket.Client, tunnelID string) chan struct{} {
+	maxInterval := 6 * time.Second
+	currentInterval := pingInterval
+	consecutiveFailures := 0
+	connectionLost := false
+
+	// Track recent latencies for adaptive timeout calculation
+	recentLatencies := make([]time.Duration, 0, 10)
+
+	pingStopChan := make(chan struct{})
+
+	go func() {
+		ticker := time.NewTicker(currentInterval)
+		defer ticker.Stop()
+		for {
+			select {
+			case <-ticker.C:
+				// Calculate adaptive timeout based on recent latencies
+				adaptiveTimeout := pingTimeout
+				if len(recentLatencies) > 0 {
+					var sum time.Duration
+					for _, lat := range recentLatencies {
+						sum += lat
+					}
+					avgLatency := sum / time.Duration(len(recentLatencies))
+					// Use 3x average latency as timeout, with minimum of pingTimeout
+					adaptiveTimeout = avgLatency * 3
+					if adaptiveTimeout < pingTimeout {
+						adaptiveTimeout = pingTimeout
+					}
+					if adaptiveTimeout > 15*time.Second {
+						adaptiveTimeout = 15 * time.Second
+					}
+				}
+
+				// Use reliable ping with multiple attempts
+				maxAttempts := 2
+				if consecutiveFailures > 4 {
+					maxAttempts = 4 // More attempts when connection is unstable
+				}
+
+				latency, err := reliablePing(tnet, serverIP, adaptiveTimeout, maxAttempts)
+				if err != nil {
+					consecutiveFailures++
+
+					// Track recent latencies (add a high value for failures)
+					recentLatencies = append(recentLatencies, adaptiveTimeout)
+					if len(recentLatencies) > 10 {
+						recentLatencies = recentLatencies[1:]
+					}
+
+					if consecutiveFailures < 2 {
+						logger.Debug("Periodic ping failed (%d consecutive failures): %v", consecutiveFailures, err)
+					} else {
+						logger.Warn("Periodic ping failed (%d consecutive failures): %v", consecutiveFailures, err)
+					}
+
+					// More lenient threshold for declaring connection lost under load
+					failureThreshold := 4
+					if shouldFireRecovery(consecutiveFailures, failureThreshold, connectionLost) {
+						connectionLost = true
+						logger.Warn("Connection to server lost after %d failures. Continuous reconnection attempts will be made.", consecutiveFailures)
+						if tunnelID != "" {
+							telemetry.IncReconnect(context.Background(), tunnelID, "client", telemetry.ReasonTimeout)
+						}
+						pingChainId := generateChainId()
+						pendingPingChainId = pingChainId
+						stopFunc = client.SendMessageInterval("newt/ping/request", map[string]interface{}{
+							"chainId": pingChainId,
+						}, 3*time.Second)
+						// Send registration message to the server for backward compatibility
+						bcChainId := generateChainId()
+						pendingRegisterChainId = bcChainId
+						err := client.SendMessage("newt/wg/register", map[string]interface{}{
+							"publicKey":           publicKey.String(),
+							"backwardsCompatible": true,
+							"chainId":             bcChainId,
+						})
+						if err != nil {
+							logger.Error("Failed to send registration message: %v", err)
+						}
+						if healthFile != "" {
+							err = os.Remove(healthFile)
+							if err != nil {
+								logger.Error("Failed to remove health file: %v", err)
+							}
+						}
+					}
+					// Backoff: ramp the periodic-ping interval up while we are
+					// past the failure threshold, capped at maxInterval. Kept
+					// independent of the recovery trigger above so the trigger
+					// fires on every outage regardless of pingInterval.
+					if consecutiveFailures >= failureThreshold && currentInterval < maxInterval {
+						currentInterval = time.Duration(float64(currentInterval) * 1.3)
+						if currentInterval > maxInterval {
+							currentInterval = maxInterval
+						}
+					}
+				} else {
+					// Track recent latencies
+					recentLatencies = append(recentLatencies, latency)
+					// Record tunnel latency (limit sampling to this periodic check)
+					if tunnelID != "" {
+						telemetry.ObserveTunnelLatency(context.Background(), tunnelID, "wireguard", latency.Seconds())
+					}
+					if len(recentLatencies) > 10 {
+						recentLatencies = recentLatencies[1:]
+					}
+
+					if connectionLost {
+						connectionLost = false
+						logger.Info("Connection to server restored after %d failures!", consecutiveFailures)
+						if healthFile != "" {
+							err := os.WriteFile(healthFile, []byte("ok"), 0644)
+							if err != nil {
+								logger.Warn("Failed to write health file: %v", err)
+							}
+						}
+					}
+					if currentInterval > pingInterval {
+						currentInterval = time.Duration(float64(currentInterval) * 0.9) // Slower decrease
+						if currentInterval < pingInterval {
+							currentInterval = pingInterval
+						}
+						ticker.Reset(currentInterval)
+						logger.Debug("Decreased ping check interval to %v after successful ping", currentInterval)
+					}
+					consecutiveFailures = 0
+				}
+			case <-pingStopChan:
+				logger.Info("Stopping ping check")
+				return
+			}
+		}
+	}()
+
+	return pingStopChan
+}
+
+func parseTargetData(data interface{}) (TargetData, error) {
+	var targetData TargetData
+	jsonData, err := json.Marshal(data)
+	if err != nil {
+		logger.Info("Error marshaling data: %v", err)
+		return targetData, err
+	}
+
+	if err := json.Unmarshal(jsonData, &targetData); err != nil {
+		logger.Info("Error unmarshaling target data: %v", err)
+		return targetData, err
+	}
+	return targetData, nil
+}
+
+// parseTargetString parses a target string in the format "listenPort:host:targetPort"
+// It properly handles IPv6 addresses which must be in brackets: "listenPort:[ipv6]:targetPort"
+// Examples:
+//   - IPv4: "3001:192.168.1.1:80"
+//   - IPv6: "3001:[::1]:8080" or "3001:[fd70:1452:b736:4dd5:caca:7db9:c588:f5b3]:80"
+//
+// Returns listenPort, targetAddress (in host:port format suitable for net.Dial), and error
+func parseTargetString(target string) (int, string, error) {
+	// Find the first colon to extract the listen port
+	firstColon := strings.Index(target, ":")
+	if firstColon == -1 {
+		return 0, "", fmt.Errorf("invalid target format, no colon found: %s", target)
+	}
+
+	listenPortStr := target[:firstColon]
+	var listenPort int
+	_, err := fmt.Sscanf(listenPortStr, "%d", &listenPort)
+	if err != nil {
+		return 0, "", fmt.Errorf("invalid listen port: %s", listenPortStr)
+	}
+	if listenPort <= 0 || listenPort > 65535 {
+		return 0, "", fmt.Errorf("listen port out of range: %d", listenPort)
+	}
+
+	// The remainder is host:targetPort - use net.SplitHostPort which handles IPv6 brackets
+	remainder := target[firstColon+1:]
+	host, targetPort, err := net.SplitHostPort(remainder)
+	if err != nil {
+		return 0, "", fmt.Errorf("invalid host:port format '%s': %w", remainder, err)
+	}
+
+	// Reject empty host or target port
+	if host == "" {
+		return 0, "", fmt.Errorf("empty host in target: %s", target)
+	}
+	if targetPort == "" {
+		return 0, "", fmt.Errorf("empty target port in target: %s", target)
+	}
+
+	// Reconstruct the target address using JoinHostPort (handles IPv6 properly)
+	targetAddr := net.JoinHostPort(host, targetPort)
+
+	return listenPort, targetAddr, nil
+}
+
+func updateTargets(pm *proxy.ProxyManager, action string, tunnelIP string, proto string, targetData TargetData) error {
+	for _, t := range targetData.Targets {
+		// Parse the target string, handling both IPv4 and IPv6 addresses
+		port, target, err := parseTargetString(t)
+		if err != nil {
+			logger.Info("Invalid target format: %s (%v)", t, err)
+			continue
+		}
+
+		switch action {
+		case "add":
+			// Call updown script if provided
+			processedTarget := target
+			if updownScript != "" {
+				newTarget, err := executeUpdownScript(action, proto, target)
+				if err != nil {
+					logger.Warn("Updown script error: %v", err)
+				} else if newTarget != "" {
+					processedTarget = newTarget
+				}
+			}
+
+			// Only remove the specific target if it exists
+			err := pm.RemoveTarget(proto, tunnelIP, port)
+			if err != nil {
+				// Ignore "target not found" errors as this is expected for new targets
+				if !strings.Contains(err.Error(), "target not found") {
+					logger.Error("Failed to remove existing target: %v", err)
+				}
+			}
+
+			// Add the new target
+			pm.AddTarget(proto, tunnelIP, port, processedTarget)
+
+		case "remove":
+			logger.Info("Removing target with port %d", port)
+
+			// Call updown script if provided
+			if updownScript != "" {
+				_, err := executeUpdownScript(action, proto, target)
+				if err != nil {
+					logger.Warn("Updown script error: %v", err)
+				}
+			}
+
+			err = pm.RemoveTarget(proto, tunnelIP, port)
+			if err != nil {
+				logger.Error("Failed to remove target: %v", err)
+				return err
+			}
+		default:
+			logger.Info("Unknown action: %s", action)
+		}
+	}
+
+	return nil
+}
+
+func executeUpdownScript(action, proto, target string) (string, error) {
+	if updownScript == "" {
+		return target, nil
+	}
+
+	// Split the updownScript in case it contains spaces (like "/usr/bin/python3 script.py")
+	parts := strings.Fields(updownScript)
+	if len(parts) == 0 {
+		return target, fmt.Errorf("invalid updown script command")
+	}
+
+	var cmd *exec.Cmd
+	if len(parts) == 1 {
+		// If it's a single executable
+		logger.Info("Executing updown script: %s %s %s %s", updownScript, action, proto, target)
+		cmd = exec.Command(parts[0], action, proto, target)
+	} else {
+		// If it includes interpreter and script
+		args := append(parts[1:], action, proto, target)
+		logger.Info("Executing updown script: %s %s %s %s %s", parts[0], strings.Join(parts[1:], " "), action, proto, target)
+		cmd = exec.Command(parts[0], args...)
+	}
+
+	output, err := cmd.Output()
+	if err != nil {
+		if exitErr, ok := err.(*exec.ExitError); ok {
+			return "", fmt.Errorf("updown script execution failed (exit code %d): %s",
+				exitErr.ExitCode(), string(exitErr.Stderr))
+		}
+		return "", fmt.Errorf("updown script execution failed: %v", err)
+	}
+
+	// If the script returns a new target, use it
+	newTarget := strings.TrimSpace(string(output))
+	if newTarget != "" {
+		logger.Info("Updown script returned new target: %s", newTarget)
+		return newTarget, nil
+	}
+
+	return target, nil
+}
+
+// interpolateBlueprint finds all {{...}} tokens in the raw blueprint bytes and
+// replaces recognised schemes with their resolved values. Currently supported:
+//
+//   - env.<VAR>  – replaced with the value of the named environment variable
+//
+// Any token that does not match a supported scheme is left as-is so that
+// future schemes (e.g. tag., api.) are preserved rather than silently dropped.
+func interpolateBlueprint(data []byte) []byte {
+	re := regexp.MustCompile(`\{\{([^}]+)\}\}`)
+	return re.ReplaceAllFunc(data, func(match []byte) []byte {
+		// strip the surrounding {{ }}
+		inner := strings.TrimSpace(string(match[2 : len(match)-2]))
+
+		if strings.HasPrefix(inner, "env.") {
+			varName := strings.TrimPrefix(inner, "env.")
+			return []byte(os.Getenv(varName))
+		}
+
+		// unrecognised scheme – leave the token untouched
+		return match
+	})
+}
+
+func sendBlueprint(client *websocket.Client, file string) error {
+	if file == "" {
+		return nil
+	}
+	// try to read the blueprint file
+	blueprintData, err := os.ReadFile(file)
+	if err != nil {
+		logger.Error("Failed to read blueprint file: %v", err)
+	} else {
+		// interpolate {{env.VAR}} (and any future schemes) before parsing
+		blueprintData = interpolateBlueprint(blueprintData)
+
+		// first we should convert the yaml to json and error if the yaml is bad
+		var yamlObj interface{}
+		var blueprintJsonData string
+
+		err = yaml.Unmarshal(blueprintData, &yamlObj)
+		if err != nil {
+			logger.Error("Failed to parse blueprint YAML: %v", err)
+		} else {
+			// convert to json
+			jsonBytes, err := json.Marshal(yamlObj)
+			if err != nil {
+				logger.Error("Failed to convert blueprint to JSON: %v", err)
+			} else {
+				blueprintJsonData = string(jsonBytes)
+				logger.Debug("Converted blueprint to JSON: %s", blueprintJsonData)
+			}
+		}
+
+		// if we have valid json data, we can send it to the server
+		if blueprintJsonData == "" {
+			logger.Error("No valid blueprint JSON data to send to server")
+			return nil
+		}
+
+		logger.Info("Sending blueprint to server for application")
+
+		// send the blueprint data to the server
+		err = client.SendMessage("newt/blueprint/apply", map[string]interface{}{
+			"blueprint": blueprintJsonData,
+		})
+	}
+
+	return nil
+}
+
+func watchBlueprintFile(ctx context.Context, filePath string, send func() error) {
+	watcher, err := fsnotify.NewWatcher()
+	if err != nil {
+		logger.Error("blueprint watcher: failed to create: %v", err)
+		return
+	}
+	defer watcher.Close()
+
+	if err := watcher.Add(filePath); err != nil {
+		logger.Error("blueprint watcher: failed to watch %s: %v", filePath, err)
+		return
+	}
+
+	logger.Info("Watching blueprint file for changes: %s", filePath)
+
+	var debounce *time.Timer
+	scheduleSend := func() {
+		if debounce != nil {
+			debounce.Stop()
+		}
+		debounce = time.AfterFunc(500*time.Millisecond, func() {
+			logger.Info("Blueprint file changed, resending...")
+			if err := send(); err != nil {
+				logger.Error("blueprint watcher: resend failed: %v", err)
+			}
+		})
+	}
+
+	for {
+		select {
+		case <-ctx.Done():
+			if debounce != nil {
+				debounce.Stop()
+			}
+			return
+		case event, ok := <-watcher.Events:
+			if !ok {
+				return
+			}
+			switch {
+			case event.Has(fsnotify.Write) || event.Has(fsnotify.Create):
+				if event.Has(fsnotify.Create) {
+					_ = watcher.Add(filePath)
+				}
+				scheduleSend()
+			case event.Has(fsnotify.Remove) || event.Has(fsnotify.Rename):
+				_ = watcher.Add(filePath)
+				scheduleSend()
+			}
+		case err, ok := <-watcher.Errors:
+			if !ok {
+				return
+			}
+			logger.Error("blueprint watcher: %v", err)
+		}
+	}
+}

common_test.go

@@ -0,0 +1,383 @@
+package main
+
+import (
+	"context"
+	"net"
+	"os"
+	"path/filepath"
+	"sync/atomic"
+	"testing"
+	"time"
+)
+
+func TestWatchBlueprintFile_WriteTriggersSend(t *testing.T) {
+	f, err := os.CreateTemp(t.TempDir(), "blueprint-*.yaml")
+	if err != nil {
+		t.Fatal(err)
+	}
+	f.Close()
+
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	var calls atomic.Int32
+	go watchBlueprintFile(ctx, f.Name(), func() error {
+		calls.Add(1)
+		return nil
+	})
+
+	time.Sleep(50 * time.Millisecond)
+
+	if err := os.WriteFile(f.Name(), []byte("content"), 0644); err != nil {
+		t.Fatal(err)
+	}
+
+	time.Sleep(700 * time.Millisecond)
+
+	if calls.Load() != 1 {
+		t.Errorf("expected 1 send call, got %d", calls.Load())
+	}
+}
+
+func TestWatchBlueprintFile_DebounceCoalescesEvents(t *testing.T) {
+	f, err := os.CreateTemp(t.TempDir(), "blueprint-*.yaml")
+	if err != nil {
+		t.Fatal(err)
+	}
+	f.Close()
+
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	var calls atomic.Int32
+	go watchBlueprintFile(ctx, f.Name(), func() error {
+		calls.Add(1)
+		return nil
+	})
+
+	time.Sleep(50 * time.Millisecond)
+
+	for i := 0; i < 5; i++ {
+		if err := os.WriteFile(f.Name(), []byte("change"), 0644); err != nil {
+			t.Fatal(err)
+		}
+		time.Sleep(50 * time.Millisecond)
+	}
+
+	time.Sleep(700 * time.Millisecond)
+
+	if calls.Load() != 1 {
+		t.Errorf("expected 1 send call after debounce, got %d", calls.Load())
+	}
+}
+
+func TestWatchBlueprintFile_ContextCancellationStops(t *testing.T) {
+	f, err := os.CreateTemp(t.TempDir(), "blueprint-*.yaml")
+	if err != nil {
+		t.Fatal(err)
+	}
+	f.Close()
+
+	ctx, cancel := context.WithCancel(context.Background())
+
+	done := make(chan struct{})
+	go func() {
+		watchBlueprintFile(ctx, f.Name(), func() error { return nil })
+		close(done)
+	}()
+
+	time.Sleep(50 * time.Millisecond)
+	cancel()
+
+	select {
+	case <-done:
+	case <-time.After(2 * time.Second):
+		t.Error("watchBlueprintFile did not exit after context cancellation")
+	}
+}
+
+func TestWatchBlueprintFile_AtomicWriteTriggersSend(t *testing.T) {
+	dir := t.TempDir()
+	target := filepath.Join(dir, "blueprint.yaml")
+	if err := os.WriteFile(target, []byte("initial"), 0644); err != nil {
+		t.Fatal(err)
+	}
+
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	var calls atomic.Int32
+	go watchBlueprintFile(ctx, target, func() error {
+		calls.Add(1)
+		return nil
+	})
+
+	time.Sleep(50 * time.Millisecond)
+
+	tmp := filepath.Join(dir, "blueprint.yaml.tmp")
+	if err := os.WriteFile(tmp, []byte("updated"), 0644); err != nil {
+		t.Fatal(err)
+	}
+	if err := os.Rename(tmp, target); err != nil {
+		t.Fatal(err)
+	}
+
+	time.Sleep(700 * time.Millisecond)
+
+	if calls.Load() < 1 {
+		t.Errorf("expected at least 1 send call after atomic write, got %d", calls.Load())
+	}
+}
+
+func TestWatchBlueprintFile_MissingFileReturnsGracefully(t *testing.T) {
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	done := make(chan struct{})
+	go func() {
+		watchBlueprintFile(ctx, "/nonexistent/path/blueprint.yaml", func() error { return nil })
+		close(done)
+	}()
+
+	select {
+	case <-done:
+	case <-time.After(2 * time.Second):
+		t.Error("watchBlueprintFile did not return for missing file")
+	}
+}
+
+func TestParseTargetString(t *testing.T) {
+	tests := []struct {
+		name           string
+		input          string
+		wantListenPort int
+		wantTargetAddr string
+		wantErr        bool
+	}{
+		{
+			name:           "valid IPv4 basic",
+			input:          "3001:192.168.1.1:80",
+			wantListenPort: 3001,
+			wantTargetAddr: "192.168.1.1:80",
+			wantErr:        false,
+		},
+		{
+			name:           "valid IPv4 localhost",
+			input:          "8080:127.0.0.1:3000",
+			wantListenPort: 8080,
+			wantTargetAddr: "127.0.0.1:3000",
+			wantErr:        false,
+		},
+		{
+			name:           "valid IPv4 same ports",
+			input:          "443:10.0.0.1:443",
+			wantListenPort: 443,
+			wantTargetAddr: "10.0.0.1:443",
+			wantErr:        false,
+		},
+		{
+			name:           "valid IPv6 loopback",
+			input:          "3001:[::1]:8080",
+			wantListenPort: 3001,
+			wantTargetAddr: "[::1]:8080",
+			wantErr:        false,
+		},
+		{
+			name:           "valid IPv6 full address",
+			input:          "80:[fd70:1452:b736:4dd5:caca:7db9:c588:f5b3]:8080",
+			wantListenPort: 80,
+			wantTargetAddr: "[fd70:1452:b736:4dd5:caca:7db9:c588:f5b3]:8080",
+			wantErr:        false,
+		},
+		{
+			name:           "valid IPv6 link-local",
+			input:          "443:[fe80::1]:443",
+			wantListenPort: 443,
+			wantTargetAddr: "[fe80::1]:443",
+			wantErr:        false,
+		},
+		{
+			name:           "valid IPv6 all zeros compressed",
+			input:          "8000:[::]:9000",
+			wantListenPort: 8000,
+			wantTargetAddr: "[::]:9000",
+			wantErr:        false,
+		},
+		{
+			name:           "valid IPv6 mixed notation",
+			input:          "5000:[::ffff:192.168.1.1]:6000",
+			wantListenPort: 5000,
+			wantTargetAddr: "[::ffff:192.168.1.1]:6000",
+			wantErr:        false,
+		},
+		{
+			name:           "valid hostname",
+			input:          "8080:example.com:80",
+			wantListenPort: 8080,
+			wantTargetAddr: "example.com:80",
+			wantErr:        false,
+		},
+		{
+			name:           "valid hostname with subdomain",
+			input:          "443:api.example.com:8443",
+			wantListenPort: 443,
+			wantTargetAddr: "api.example.com:8443",
+			wantErr:        false,
+		},
+		{
+			name:           "valid localhost hostname",
+			input:          "3000:localhost:3000",
+			wantListenPort: 3000,
+			wantTargetAddr: "localhost:3000",
+			wantErr:        false,
+		},
+		{
+			name:    "invalid - no colons",
+			input:   "invalid",
+			wantErr: true,
+		},
+		{
+			name:    "invalid - empty string",
+			input:   "",
+			wantErr: true,
+		},
+		{
+			name:    "invalid - non-numeric listen port",
+			input:   "abc:192.168.1.1:80",
+			wantErr: true,
+		},
+		{
+			name:    "invalid - missing target port",
+			input:   "3001:192.168.1.1",
+			wantErr: true,
+		},
+		{
+			name:    "invalid - IPv6 without brackets",
+			input:   "3001:fd70:1452:b736:4dd5:caca:7db9:c588:f5b3:80",
+			wantErr: true,
+		},
+		{
+			name:    "invalid - only listen port",
+			input:   "3001:",
+			wantErr: true,
+		},
+		{
+			name:    "invalid - missing host",
+			input:   "3001::80",
+			wantErr: true,
+		},
+		{
+			name:    "invalid - IPv6 unclosed bracket",
+			input:   "3001:[::1:80",
+			wantErr: true,
+		},
+		{
+			name:    "invalid - listen port zero",
+			input:   "0:192.168.1.1:80",
+			wantErr: true,
+		},
+		{
+			name:    "invalid - listen port negative",
+			input:   "-1:192.168.1.1:80",
+			wantErr: true,
+		},
+		{
+			name:    "invalid - listen port out of range",
+			input:   "70000:192.168.1.1:80",
+			wantErr: true,
+		},
+		{
+			name:    "invalid - empty target port",
+			input:   "3001:192.168.1.1:",
+			wantErr: true,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			listenPort, targetAddr, err := parseTargetString(tt.input)
+
+			if (err != nil) != tt.wantErr {
+				t.Errorf("parseTargetString(%q) error = %v, wantErr %v", tt.input, err, tt.wantErr)
+				return
+			}
+
+			if tt.wantErr {
+				return
+			}
+
+			if listenPort != tt.wantListenPort {
+				t.Errorf("parseTargetString(%q) listenPort = %d, want %d", tt.input, listenPort, tt.wantListenPort)
+			}
+
+			if targetAddr != tt.wantTargetAddr {
+				t.Errorf("parseTargetString(%q) targetAddr = %q, want %q", tt.input, targetAddr, tt.wantTargetAddr)
+			}
+		})
+	}
+}
+
+// TestParseTargetStringNetDialCompatibility verifies that the output is compatible with net.Dial.
+func TestParseTargetStringNetDialCompatibility(t *testing.T) {
+	tests := []struct {
+		name  string
+		input string
+	}{
+		{"IPv4", "8080:127.0.0.1:80"},
+		{"IPv6 loopback", "8080:[::1]:80"},
+		{"IPv6 full", "8080:[2001:db8::1]:80"},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			_, targetAddr, err := parseTargetString(tt.input)
+			if err != nil {
+				t.Fatalf("parseTargetString(%q) unexpected error: %v", tt.input, err)
+			}
+
+			_, _, err = net.SplitHostPort(targetAddr)
+			if err != nil {
+				t.Errorf("parseTargetString(%q) produced invalid net.Dial format %q: %v", tt.input, targetAddr, err)
+			}
+		})
+	}
+}
+
+// TestShouldFireRecovery is the regression guard for the broken trigger gate
+// that prevented data-plane recovery from ever firing under default settings
+// (fosrl/newt#284, #310, pangolin#1004). The pre-fix condition was
+//
+//	consecutiveFailures >= failureThreshold && currentInterval < maxInterval
+//
+// which became permanently false once pingInterval's default was bumped from
+// 3s to 15s in commit 8161fa6 — currentInterval starts at pingInterval=15s,
+// maxInterval stayed at 6s, so 15<6 is false and the recovery branch never
+// executed.
+//
+// The fix is to drop currentInterval from the trigger condition entirely;
+// backoff is a separate concern computed in the caller. The cases below
+// exercise the documented contract.
+func TestShouldFireRecovery(t *testing.T) {
+	const threshold = 4
+	cases := []struct {
+		name           string
+		failures       int
+		connectionLost bool
+		want           bool
+	}{
+		{"below threshold, fresh", 3, false, false},
+		{"below threshold, already lost", 3, true, false},
+		{"at threshold, fresh — recovery must fire", threshold, false, true},
+		{"at threshold, already lost — gate prevents re-fire", threshold, true, false},
+		{"far above threshold, fresh", 100, false, true},
+		{"far above threshold, already lost", 100, true, false},
+	}
+	for _, c := range cases {
+		t.Run(c.name, func(t *testing.T) {
+			if got := shouldFireRecovery(c.failures, threshold, c.connectionLost); got != c.want {
+				t.Errorf("shouldFireRecovery(failures=%d, threshold=%d, lost=%v) = %v, want %v",
+					c.failures, threshold, c.connectionLost, got, c.want)
+			}
+		})
+	}
+}

device/tun_unix.go

@@ -0,0 +1,44 @@
+//go:build !windows
+
+package device
+
+import (
+	"net"
+	"os"
+
+	"github.com/fosrl/newt/logger"
+	"golang.org/x/sys/unix"
+	"golang.zx2c4.com/wireguard/ipc"
+	"golang.zx2c4.com/wireguard/tun"
+)
+
+func CreateTUNFromFD(tunFd uint32, mtuInt int) (tun.Device, error) {
+	dupTunFd, err := unix.Dup(int(tunFd))
+	if err != nil {
+		logger.Error("Unable to dup tun fd: %v", err)
+		return nil, err
+	}
+
+	err = unix.SetNonblock(dupTunFd, true)
+	if err != nil {
+		unix.Close(dupTunFd)
+		return nil, err
+	}
+
+	file := os.NewFile(uintptr(dupTunFd), "/dev/tun")
+	device, err := tun.CreateTUNFromFile(file, mtuInt)
+	if err != nil {
+		file.Close()
+		return nil, err
+	}
+
+	return device, nil
+}
+
+func UapiOpen(interfaceName string) (*os.File, error) {
+	return ipc.UAPIOpen(interfaceName)
+}
+
+func UapiListen(interfaceName string, fileUAPI *os.File) (net.Listener, error) {
+	return ipc.UAPIListen(interfaceName, fileUAPI)
+}

device/tun_windows.go

@@ -0,0 +1,25 @@
+//go:build windows
+
+package device
+
+import (
+	"errors"
+	"net"
+	"os"
+
+	"golang.zx2c4.com/wireguard/ipc"
+	"golang.zx2c4.com/wireguard/tun"
+)
+
+func CreateTUNFromFD(tunFd uint32, mtuInt int) (tun.Device, error) {
+	return nil, errors.New("CreateTUNFromFile not supported on Windows")
+}
+
+func UapiOpen(interfaceName string) (*os.File, error) {
+	return nil, nil
+}
+
+func UapiListen(interfaceName string, fileUAPI *os.File) (net.Listener, error) {
+	// On Windows, UAPIListen only takes one parameter
+	return ipc.UAPIListen(interfaceName)
+}

docker-compose.metrics.collector.yml

@@ -0,0 +1,41 @@
+services:
+  newt:
+    build: .
+    image: newt:dev
+    env_file:
+      - .env
+    environment:
+      - NEWT_METRICS_PROMETHEUS_ENABLED=false   # important: disable direct /metrics scraping
+      - NEWT_METRICS_OTLP_ENABLED=true          # OTLP to the Collector
+      # optional:
+      # - NEWT_METRICS_INCLUDE_TUNNEL_ID=false
+  # When using the Collector pattern, do NOT map the Newt admin/metrics port
+  # (2112) on the application service. Mapping 2112 here can cause port
+  # conflicts and may result in duplicated Prometheus scraping (app AND
+  # collector being scraped for the same metrics). Instead either:
+  #  - leave ports unset on the app service (recommended), or
+  #  - map 2112 only on a dedicated metrics/collector service that is
+  #    responsible for exposing metrics to Prometheus.
+  # Example: do NOT map here
+  # ports: []
+  # Example: map 2112 only on a collector service
+  # collector:
+  #   ports:
+  #     - "2112:2112"  # collector's prometheus exporter (scraped by Prometheus)
+
+  otel-collector:
+    image: otel/opentelemetry-collector-contrib:latest
+    command: ["--config=/etc/otelcol/config.yaml"]
+    volumes:
+      - ./examples/otel-collector.yaml:/etc/otelcol/config.yaml:ro
+    ports:
+      - "4317:4317"   # OTLP gRPC
+      - "8889:8889"   # Prometheus Exporter (scraped by Prometheus)
+
+  prometheus:
+    image: prom/prometheus:latest
+    volumes:
+      - ./examples/prometheus.with-collector.yml:/etc/prometheus/prometheus.yml:ro
+    ports:
+      - "9090:9090"
+

docker-compose.metrics.yml

@@ -0,0 +1,56 @@
+name: Newt-Metrics
+services:
+  # Recommended Variant A: Direct Prometheus scrape of Newt (/metrics)
+  # Optional: You may add the Collector service and enable OTLP export, but do NOT
+  # scrape both Newt and the Collector for the same process.
+
+  newt:
+    build: .
+    image: newt:dev
+    env_file:
+      - .env
+    environment:
+      OTEL_SERVICE_NAME: newt
+      NEWT_METRICS_PROMETHEUS_ENABLED: "true"
+      NEWT_METRICS_OTLP_ENABLED: "false"   # avoid double-scrape by default
+      NEWT_ADMIN_ADDR: ":2112"
+      # Base NEWT configuration
+      PANGOLIN_ENDPOINT: ${PANGOLIN_ENDPOINT}
+      NEWT_ID: ${NEWT_ID}
+      NEWT_SECRET: ${NEWT_SECRET}
+      LOG_LEVEL: "DEBUG"
+    ports:
+      - "2112:2112"
+
+  # Optional Variant B: Enable the Collector and switch Prometheus scrape to it.
+  # collector:
+  #   image: otel/opentelemetry-collector-contrib:0.136.0
+  #   command: ["--config=/etc/otelcol/config.yaml"]
+  #   volumes:
+  #     - ./examples/otel-collector.yaml:/etc/otelcol/config.yaml:ro
+  #   ports:
+  #     - "4317:4317"   # OTLP gRPC in
+  #     - "8889:8889"   # Prometheus scrape out
+
+  prometheus:
+    image: prom/prometheus:v3.6.0
+    volumes:
+      - ./examples/prometheus.yml:/etc/prometheus/prometheus.yml:ro
+    ports:
+      - "9090:9090"
+
+  grafana:
+    image: grafana/grafana:12.2.0
+    container_name: newt-metrics-grafana
+    restart: unless-stopped
+    environment:
+      - GF_SECURITY_ADMIN_USER=admin
+      - GF_SECURITY_ADMIN_PASSWORD=admin
+    ports:
+      - "3005:3000"
+    depends_on:
+      - prometheus
+    volumes:
+      - ./examples/grafana/provisioning/datasources:/etc/grafana/provisioning/datasources:ro
+      - ./examples/grafana/provisioning/dashboards:/etc/grafana/provisioning/dashboards:ro
+      - ./examples/grafana/dashboards:/var/lib/grafana/dashboards:ro

docker-compose.yml

@@ -4,7 +4,7 @@ services:
     container_name: newt
     restart: unless-stopped
     environment:
-      - PANGOLIN_ENDPOINT=https://example.com
+      - PANGOLIN_ENDPOINT=https://app.pangolin.net
       - NEWT_ID=2ix2t8xk22ubpfy 
       - NEWT_SECRET=nnisrfsdfc7prqsp9ewo1dvtvci50j5uiqotez00dgap0ii2 
       - LOG_LEVEL=DEBUG

docker/docker.go

@@ -0,0 +1,457 @@
+package docker
+
+import (
+	"context"
+	"fmt"
+	"net"
+	"os"
+	"strconv"
+	"strings"
+	"time"
+
+	"github.com/moby/moby/api/types/container"
+	"github.com/moby/moby/api/types/events"
+	"github.com/moby/moby/client"
+	"github.com/fosrl/newt/logger"
+)
+
+// Container represents a Docker container
+type Container struct {
+	ID       string             `json:"id"`
+	Name     string             `json:"name"`
+	Image    string             `json:"image"`
+	State    string             `json:"state"`
+	Status   string             `json:"status"`
+	Ports    []Port             `json:"ports"`
+	Labels   map[string]string  `json:"labels"`
+	Created  int64              `json:"created"`
+	Networks map[string]Network `json:"networks"`
+	Hostname string             `json:"hostname"` // added to use hostname if available instead of network address
+
+}
+
+// Port represents a port mapping for a Docker container
+type Port struct {
+	PrivatePort int    `json:"privatePort"`
+	PublicPort  int    `json:"publicPort,omitempty"`
+	Type        string `json:"type"`
+	IP          string `json:"ip,omitempty"`
+}
+
+// Network represents network information for a Docker container
+type Network struct {
+	NetworkID           string   `json:"networkId"`
+	EndpointID          string   `json:"endpointId"`
+	Gateway             string   `json:"gateway,omitempty"`
+	IPAddress           string   `json:"ipAddress,omitempty"`
+	IPPrefixLen         int      `json:"ipPrefixLen,omitempty"`
+	IPv6Gateway         string   `json:"ipv6Gateway,omitempty"`
+	GlobalIPv6Address   string   `json:"globalIPv6Address,omitempty"`
+	GlobalIPv6PrefixLen int      `json:"globalIPv6PrefixLen,omitempty"`
+	MacAddress          string   `json:"macAddress,omitempty"`
+	Aliases             []string `json:"aliases,omitempty"`
+	DNSNames            []string `json:"dnsNames,omitempty"`
+}
+
+// Strcuture parts of docker api endpoint
+type dockerHost struct {
+	protocol string // e.g. unix, http, tcp, ssh
+	address  string // e.g. "/var/run/docker.sock" or "host:port"
+}
+
+// Parse the docker api endpoint into its parts
+func parseDockerHost(raw string) (dockerHost, error) {
+	switch {
+	case strings.HasPrefix(raw, "unix://"):
+		return dockerHost{"unix", strings.TrimPrefix(raw, "unix://")}, nil
+	case strings.HasPrefix(raw, "ssh://"):
+		// SSH is treated as TCP-like transport by the docker client
+		return dockerHost{"ssh", strings.TrimPrefix(raw, "ssh://")}, nil
+	case strings.HasPrefix(raw, "tcp://"), strings.HasPrefix(raw, "http://"), strings.HasPrefix(raw, "https://"):
+		s := raw
+		s = strings.TrimPrefix(s, "tcp://")
+		s = strings.TrimPrefix(s, "http://")
+		s = strings.TrimPrefix(s, "https://")
+		return dockerHost{"tcp", s}, nil
+	case strings.HasPrefix(raw, "/"):
+		// Absolute path without scheme - treat as unix socket
+		return dockerHost{"unix", raw}, nil
+	default:
+		// For relative paths or other formats, also default to unix
+		return dockerHost{"unix", raw}, nil
+	}
+}
+
+// CheckSocket checks if Docker socket is available
+func CheckSocket(socketPath string) bool {
+	// Use the provided socket path or default to standard location
+	if socketPath == "" {
+		socketPath = "unix:///var/run/docker.sock"
+	}
+
+	// Ensure the socket path is properly formatted
+	if !strings.Contains(socketPath, "://") {
+		// If no scheme provided, assume unix socket
+		socketPath = "unix://" + socketPath
+	}
+
+	host, err := parseDockerHost(socketPath)
+	if err != nil {
+		logger.Debug("Invalid Docker socket path '%s': %v", socketPath, err)
+		return false
+	}
+	protocol := host.protocol
+	addr := host.address
+
+	// ssh might need different verification, but tcp works for basic reachability
+	conn, err := net.DialTimeout(protocol, addr, 2*time.Second)
+	if err != nil {
+		logger.Debug("Docker not reachable via %s at %s: %v", protocol, addr, err)
+		return false
+	}
+	defer conn.Close()
+
+	logger.Debug("Docker reachable via %s at %s", protocol, addr)
+	return true
+}
+
+// IsWithinHostNetwork checks if a provided target is within the host container network
+func IsWithinHostNetwork(socketPath string, targetAddress string, targetPort int) (bool, error) {
+	// Always enforce network validation
+	containers, err := ListContainers(socketPath, true)
+	if err != nil {
+		return false, err
+	}
+
+	// Determine if given an IP address
+	var parsedTargetAddressIp = net.ParseIP(targetAddress)
+
+	// If we can find the passed hostname/IP address in the networks or as the container name, it is valid and can add it
+	for _, c := range containers {
+		for _, network := range c.Networks {
+			// If the target address is not an IP address, use the container name
+			if parsedTargetAddressIp == nil {
+				if c.Name == targetAddress {
+					for _, port := range c.Ports {
+						if port.PublicPort == targetPort || port.PrivatePort == targetPort {
+							return true, nil
+						}
+					}
+				}
+			} else {
+				//If the IP address matches, check the ports being mapped too
+				if network.IPAddress == targetAddress {
+					for _, port := range c.Ports {
+						if port.PublicPort == targetPort || port.PrivatePort == targetPort {
+							return true, nil
+						}
+					}
+				}
+			}
+		}
+	}
+
+	combinedTargetAddress := targetAddress + ":" + strconv.Itoa(targetPort)
+	return false, fmt.Errorf("target address not within host container network: %s", combinedTargetAddress)
+}
+
+// ListContainers lists all Docker containers with their network information
+func ListContainers(socketPath string, enforceNetworkValidation bool) ([]Container, error) {
+	// Use the provided socket path or default to standard location
+	if socketPath == "" {
+		socketPath = "unix:///var/run/docker.sock"
+	}
+
+	// Ensure the socket path is properly formatted for the Docker client
+	if !strings.Contains(socketPath, "://") {
+		// If no scheme provided, assume unix socket
+		socketPath = "unix://" + socketPath
+	}
+
+	// Used to filter down containers returned to Pangolin
+	containerFilters := make(client.Filters)
+
+	// Used to determine if we will send IP addresses or hostnames to Pangolin
+	useContainerIpAddresses := true
+	hostContainerId := ""
+
+	// Create a new Docker client
+	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+	defer cancel()
+
+	// Create client with custom socket path
+	cli, err := client.NewClientWithOpts(
+		client.WithHost(socketPath),
+		client.WithAPIVersionNegotiation(),
+	)
+	if err != nil {
+		return nil, fmt.Errorf("failed to create Docker client: %v", err)
+	}
+
+	defer cli.Close()
+
+	hostContainer, err := getHostContainer(ctx, cli)
+	if enforceNetworkValidation && err != nil {
+		return nil, fmt.Errorf("network validation enforced, cannot validate due to: %w", err)
+	}
+
+	// We may not be able to get back host container in scenarios like running the container in network mode 'host'
+	if hostContainer != nil {
+		// We can use the host container to filter out the list of returned containers
+		hostContainerId = hostContainer.ID
+
+		for hostContainerNetworkName := range hostContainer.NetworkSettings.Networks {
+			// If we're enforcing network validation, we'll filter on the host containers networks
+			if enforceNetworkValidation {
+				containerFilters.Add("network", hostContainerNetworkName)
+			}
+
+			// If the container is on the docker bridge network, we will use IP addresses over hostnames
+			if useContainerIpAddresses && hostContainerNetworkName != "bridge" {
+				useContainerIpAddresses = false
+			}
+		}
+	}
+
+	// List containers
+	containerListResult, err := cli.ContainerList(ctx, client.ContainerListOptions{All: true, Filters: containerFilters})
+	if err != nil {
+		return nil, fmt.Errorf("failed to list containers: %v", err)
+	}
+
+	addrToString := func(addr interface{ IsValid() bool; String() string }) string {
+		if addr.IsValid() {
+			return addr.String()
+		}
+		return ""
+	}
+
+	var dockerContainers []Container
+	for _, c := range containerListResult.Items {
+		// Short ID like docker ps
+		shortId := c.ID[:12]
+
+		// Inspect container to get hostname
+		hostname := ""
+		containerInfo, err := cli.ContainerInspect(ctx, c.ID, client.ContainerInspectOptions{})
+		if err == nil && containerInfo.Container.Config != nil {
+			hostname = containerInfo.Container.Config.Hostname
+		}
+
+		// Skip host container if set
+		if hostContainerId != "" && c.ID == hostContainerId {
+			continue
+		}
+
+		// Get container name (remove leading slash)
+		name := ""
+		if len(c.Names) > 0 {
+			name = strings.TrimPrefix(c.Names[0], "/")
+		}
+
+		// Convert ports
+		var ports []Port
+		for _, port := range c.Ports {
+			dockerPort := Port{
+				PrivatePort: int(port.PrivatePort),
+				Type:        port.Type,
+			}
+			if port.PublicPort != 0 {
+				dockerPort.PublicPort = int(port.PublicPort)
+			}
+			if port.IP.IsValid() {
+				dockerPort.IP = port.IP.String()
+			}
+			ports = append(ports, dockerPort)
+		}
+
+		// Get network information by inspecting the container
+		networks := make(map[string]Network)
+
+		// Extract network information from inspection
+		if c.NetworkSettings != nil && c.NetworkSettings.Networks != nil {
+			for networkName, endpoint := range c.NetworkSettings.Networks {
+				dockerNetwork := Network{
+					NetworkID:           endpoint.NetworkID,
+					EndpointID:          endpoint.EndpointID,
+					Gateway:             addrToString(endpoint.Gateway),
+					IPPrefixLen:         endpoint.IPPrefixLen,
+					IPv6Gateway:         addrToString(endpoint.IPv6Gateway),
+					GlobalIPv6Address:   addrToString(endpoint.GlobalIPv6Address),
+					GlobalIPv6PrefixLen: endpoint.GlobalIPv6PrefixLen,
+					MacAddress:          endpoint.MacAddress.String(),
+					Aliases:             endpoint.Aliases,
+					DNSNames:            endpoint.DNSNames,
+				}
+
+				// Use IPs over hostnames/containers as we're on the bridge network
+				if useContainerIpAddresses {
+					dockerNetwork.IPAddress = addrToString(endpoint.IPAddress)
+				}
+
+				networks[networkName] = dockerNetwork
+			}
+		}
+
+		dockerContainer := Container{
+			ID:       shortId,
+			Name:     name,
+			Image:    c.Image,
+			State:    string(c.State),
+			Status:   c.Status,
+			Ports:    ports,
+			Labels:   c.Labels,
+			Created:  c.Created,
+			Networks: networks,
+			Hostname: hostname, // added
+		}
+
+		dockerContainers = append(dockerContainers, dockerContainer)
+	}
+
+	return dockerContainers, nil
+}
+
+// getHostContainer gets the current container for the current host if possible
+func getHostContainer(dockerContext context.Context, dockerClient *client.Client) (*container.InspectResponse, error) {
+	// Get hostname from the os
+	hostContainerName, err := os.Hostname()
+	if err != nil {
+		return nil, fmt.Errorf("failed to find hostname for container")
+	}
+
+	// Get host container from the docker socket
+	hostContainer, err := dockerClient.ContainerInspect(dockerContext, hostContainerName, client.ContainerInspectOptions{})
+	if err != nil {
+		return nil, fmt.Errorf("failed to find host container")
+	}
+
+	return &hostContainer.Container, nil
+}
+
+// EventCallback defines the function signature for handling Docker events
+type EventCallback func(containers []Container)
+
+// EventMonitor handles Docker event monitoring
+type EventMonitor struct {
+	client                   *client.Client
+	ctx                      context.Context
+	cancel                   context.CancelFunc
+	callback                 EventCallback
+	socketPath               string
+	enforceNetworkValidation bool
+}
+
+// NewEventMonitor creates a new Docker event monitor
+func NewEventMonitor(socketPath string, enforceNetworkValidation bool, callback EventCallback) (*EventMonitor, error) {
+	if socketPath == "" {
+		socketPath = "unix:///var/run/docker.sock"
+	}
+
+	if !strings.Contains(socketPath, "://") {
+		socketPath = "unix://" + socketPath
+	}
+
+	cli, err := client.NewClientWithOpts(
+		client.WithHost(socketPath),
+		client.WithAPIVersionNegotiation(),
+	)
+	if err != nil {
+		return nil, fmt.Errorf("failed to create Docker client: %v", err)
+	}
+
+	ctx, cancel := context.WithCancel(context.Background())
+
+	return &EventMonitor{
+		client:                   cli,
+		ctx:                      ctx,
+		cancel:                   cancel,
+		callback:                 callback,
+		socketPath:               socketPath,
+		enforceNetworkValidation: enforceNetworkValidation,
+	}, nil
+}
+
+// Start begins monitoring Docker events
+func (em *EventMonitor) Start() error {
+	logger.Debug("Starting Docker event monitoring")
+
+	// Filter for container events we care about
+	eventFilters := make(client.Filters)
+	eventFilters.Add("type", "container")
+	// eventFilters.Add("event", "create")
+	eventFilters.Add("event", "start")
+	eventFilters.Add("event", "stop")
+	// eventFilters.Add("event", "destroy")
+	// eventFilters.Add("event", "die")
+	// eventFilters.Add("event", "pause")
+	// eventFilters.Add("event", "unpause")
+
+	// Start listening for events
+	eventsResult := em.client.Events(em.ctx, client.EventsListOptions{
+		Filters: eventFilters,
+	})
+	eventCh, errCh := eventsResult.Messages, eventsResult.Err
+
+	go func() {
+		defer func() {
+			if err := em.client.Close(); err != nil {
+				logger.Error("Error closing Docker client: %v", err)
+			}
+		}()
+
+		for {
+			select {
+			case event := <-eventCh:
+				logger.Debug("Docker event received: %s %s for container %s", event.Action, event.Type, event.Actor.ID[:12])
+
+				// Fetch updated container list and trigger callback
+				go em.handleEvent(event)
+
+			case err := <-errCh:
+				if err != nil && err != context.Canceled {
+					logger.Error("Docker event stream error: %v", err)
+					// Try to reconnect after a brief delay
+					time.Sleep(5 * time.Second)
+					if em.ctx.Err() == nil {
+						logger.Info("Attempting to reconnect to Docker event stream")
+						eventsResult = em.client.Events(em.ctx, client.EventsListOptions{
+							Filters: eventFilters,
+						})
+						eventCh, errCh = eventsResult.Messages, eventsResult.Err
+					}
+				}
+				return
+
+			case <-em.ctx.Done():
+				logger.Info("Docker event monitoring stopped")
+				return
+			}
+		}
+	}()
+
+	return nil
+}
+
+// handleEvent processes a Docker event and triggers the callback with updated container list
+func (em *EventMonitor) handleEvent(event events.Message) {
+	// Add a small delay to ensure Docker has fully processed the event
+	time.Sleep(100 * time.Millisecond)
+
+	containers, err := ListContainers(em.socketPath, em.enforceNetworkValidation)
+	if err != nil {
+		logger.Error("Failed to list containers after Docker event %s: %v", event.Action, err)
+		return
+	}
+
+	logger.Debug("Triggering callback with %d containers after Docker event %s", len(containers), event.Action)
+	em.callback(containers)
+}
+
+// Stop stops the event monitoring
+func (em *EventMonitor) Stop() {
+	logger.Info("Stopping Docker event monitoring")
+	if em.cancel != nil {
+		em.cancel()
+	}
+}

examples/README.md

@@ -0,0 +1,167 @@
+# Extensible Logger
+
+This logger package provides a flexible logging system that can be extended with custom log writers.
+
+## Basic Usage (Current Behavior)
+
+The logger works exactly as before with no changes required:
+
+```go
+package main
+
+import "your-project/logger"
+
+func main() {
+    // Use default logger
+    logger.Info("This works as before")
+    logger.Debug("Debug message")
+    logger.Error("Error message")
+    
+    // Or create a custom instance
+    log := logger.NewLogger()
+    log.SetLevel(logger.INFO)
+    log.Info("Custom logger instance")
+}
+```
+
+## Custom Log Writers
+
+To use a custom log backend, implement the `LogWriter` interface:
+
+```go
+type LogWriter interface {
+    Write(level LogLevel, timestamp time.Time, message string)
+}
+```
+
+### Example: OS Log Writer (macOS/iOS)
+
+```go
+package main
+
+import "your-project/logger"
+
+func main() {
+    // Create an OS log writer
+    osWriter := logger.NewOSLogWriter(
+        "net.pangolin.Pangolin.PacketTunnel",
+        "PangolinGo",
+        "MyApp",
+    )
+    
+    // Create a logger with the OS log writer
+    log := logger.NewLoggerWithWriter(osWriter)
+    log.SetLevel(logger.DEBUG)
+    
+    // Use it just like the standard logger
+    log.Info("This message goes to os_log")
+    log.Error("Error logged to os_log")
+}
+```
+
+### Example: Custom Writer
+
+```go
+package main
+
+import (
+    "fmt"
+    "time"
+    "your-project/logger"
+)
+
+// CustomWriter writes logs to a custom destination
+type CustomWriter struct {
+    // your custom fields
+}
+
+func (w *CustomWriter) Write(level logger.LogLevel, timestamp time.Time, message string) {
+    // Your custom logging logic
+    fmt.Printf("[CUSTOM] %s [%s] %s\n", timestamp.Format(time.RFC3339), level.String(), message)
+}
+
+func main() {
+    customWriter := &CustomWriter{}
+    log := logger.NewLoggerWithWriter(customWriter)
+    log.Info("Custom logging!")
+}
+```
+
+### Example: Multi-Writer (Log to Multiple Destinations)
+
+```go
+package main
+
+import (
+    "time"
+    "your-project/logger"
+)
+
+// MultiWriter writes to multiple log writers
+type MultiWriter struct {
+    writers []logger.LogWriter
+}
+
+func NewMultiWriter(writers ...logger.LogWriter) *MultiWriter {
+    return &MultiWriter{writers: writers}
+}
+
+func (w *MultiWriter) Write(level logger.LogLevel, timestamp time.Time, message string) {
+    for _, writer := range w.writers {
+        writer.Write(level, timestamp, message)
+    }
+}
+
+func main() {
+    // Log to both standard output and OS log
+    standardWriter := logger.NewStandardWriter()
+    osWriter := logger.NewOSLogWriter("com.example.app", "Main", "App")
+    
+    multiWriter := NewMultiWriter(standardWriter, osWriter)
+    log := logger.NewLoggerWithWriter(multiWriter)
+    
+    log.Info("This goes to both stdout and os_log!")
+}
+```
+
+## API Reference
+
+### Creating Loggers
+
+- `NewLogger()` - Creates a logger with the default StandardWriter
+- `NewLoggerWithWriter(writer LogWriter)` - Creates a logger with a custom writer
+
+### Built-in Writers
+
+- `NewStandardWriter()` - Standard writer that outputs to stdout (default)
+- `NewOSLogWriter(subsystem, category, prefix string)` - OS log writer for macOS/iOS (example)
+
+### Logger Methods
+
+- `SetLevel(level LogLevel)` - Set minimum log level
+- `SetOutput(output *os.File)` - Set output file (StandardWriter only)
+- `Debug(format string, args ...interface{})` - Log debug message
+- `Info(format string, args ...interface{})` - Log info message
+- `Warn(format string, args ...interface{})` - Log warning message
+- `Error(format string, args ...interface{})` - Log error message
+- `Fatal(format string, args ...interface{})` - Log fatal message and exit
+
+### Global Functions
+
+For convenience, you can use global functions that use the default logger:
+
+- `logger.Debug(format, args...)`
+- `logger.Info(format, args...)`
+- `logger.Warn(format, args...)`
+- `logger.Error(format, args...)`
+- `logger.Fatal(format, args...)`
+- `logger.SetOutput(output *os.File)`
+
+## Migration Guide
+
+No changes needed! The logger maintains 100% backward compatibility. Your existing code will continue to work without modifications.
+
+If you want to switch to a custom writer:
+1. Create your writer implementing `LogWriter`
+2. Use `NewLoggerWithWriter()` instead of `NewLogger()`
+3. That's it!

examples/grafana/dashboards/newt-overview.json

@@ -0,0 +1,898 @@
+{
+  "annotations": {
+    "list": [
+      {
+        "builtIn": 1,
+        "datasource": {
+          "type": "grafana",
+          "uid": "-- Grafana --"
+        },
+        "enable": true,
+        "hide": true,
+        "iconColor": "rgba(0, 211, 255, 1)",
+        "name": "Annotations & Alerts",
+        "type": "dashboard"
+      }
+    ]
+  },
+  "editable": true,
+  "fiscalYearStartMonth": 0,
+  "graphTooltip": 0,
+  "id": null,
+  "links": [],
+  "liveNow": false,
+  "panels": [
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "prometheus"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "thresholds"
+          },
+          "decimals": 0,
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green",
+                "value": null
+              },
+              {
+                "color": "red",
+                "value": 500
+              }
+            ]
+          }
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 7,
+        "w": 6,
+        "x": 0,
+        "y": 0
+      },
+      "id": 1,
+      "options": {
+        "colorMode": "value",
+        "graphMode": "area",
+        "justifyMode": "auto",
+        "orientation": "horizontal",
+        "reduceOptions": {
+          "calcs": [
+            "lastNotNull"
+          ],
+          "fields": "",
+          "values": false
+        },
+        "textMode": "value_and_name"
+      },
+      "pluginVersion": "11.1.0",
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "prometheus"
+          },
+          "editorMode": "code",
+          "expr": "go_goroutine_count",
+          "instant": true,
+          "legendFormat": "",
+          "refId": "A"
+        }
+      ],
+      "title": "Goroutines",
+      "transformations": [],
+      "type": "stat"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "prometheus"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "thresholds"
+          },
+          "decimals": 1,
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green",
+                "value": null
+              },
+              {
+                "color": "orange",
+                "value": 256
+              },
+              {
+                "color": "red",
+                "value": 512
+              }
+            ]
+          },
+          "unit": "bytes"
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 7,
+        "w": 6,
+        "x": 6,
+        "y": 0
+      },
+      "id": 2,
+      "options": {
+        "colorMode": "background",
+        "graphMode": "area",
+        "justifyMode": "auto",
+        "orientation": "horizontal",
+        "reduceOptions": {
+          "calcs": [
+            "lastNotNull"
+          ],
+          "fields": "",
+          "values": false
+        },
+        "textMode": "value_and_name"
+      },
+      "pluginVersion": "11.1.0",
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "prometheus"
+          },
+          "editorMode": "code",
+          "expr": "go_memory_gc_goal_bytes / 1024 / 1024",
+          "format": "time_series",
+          "instant": true,
+          "legendFormat": "",
+          "refId": "A"
+        }
+      ],
+      "title": "GC Target Heap (MiB)",
+      "transformations": [],
+      "type": "stat"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "prometheus"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "thresholds"
+          },
+          "decimals": 2,
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green",
+                "value": null
+              },
+              {
+                "color": "orange",
+                "value": 10
+              },
+              {
+                "color": "red",
+                "value": 25
+              }
+            ]
+          },
+          "unit": "ops"
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 7,
+        "w": 6,
+        "x": 12,
+        "y": 0
+      },
+      "id": 3,
+      "options": {
+        "colorMode": "value",
+        "graphMode": "area",
+        "justifyMode": "auto",
+        "orientation": "horizontal",
+        "reduceOptions": {
+          "calcs": [
+            "lastNotNull"
+          ],
+          "fields": "",
+          "values": false
+        },
+        "textMode": "value_and_name"
+      },
+      "pluginVersion": "11.1.0",
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "prometheus"
+          },
+          "editorMode": "code",
+          "expr": "sum(rate(http_server_request_duration_seconds_count[$__rate_interval]))",
+          "instant": false,
+          "legendFormat": "req/s",
+          "refId": "A"
+        }
+      ],
+      "title": "HTTP Requests / s",
+      "transformations": [],
+      "type": "stat"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "prometheus"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "color": {
+            "mode": "thresholds"
+          },
+          "decimals": 3,
+          "mappings": [],
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              {
+                "color": "green",
+                "value": null
+              },
+              {
+                "color": "orange",
+                "value": 0.1
+              },
+              {
+                "color": "red",
+                "value": 0.5
+              }
+            ]
+          },
+          "unit": "ops"
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 7,
+        "w": 6,
+        "x": 18,
+        "y": 0
+      },
+      "id": 4,
+      "options": {
+        "colorMode": "value",
+        "graphMode": "area",
+        "justifyMode": "auto",
+        "orientation": "horizontal",
+        "reduceOptions": {
+          "calcs": [
+            "lastNotNull"
+          ],
+          "fields": "",
+          "values": false
+        },
+        "textMode": "value_and_name"
+      },
+      "pluginVersion": "11.1.0",
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "prometheus"
+          },
+          "editorMode": "code",
+          "expr": "sum(rate(newt_connection_errors_total{site_id=~\"$site_id\"}[$__rate_interval]))",
+          "instant": false,
+          "legendFormat": "errors/s",
+          "refId": "A"
+        }
+      ],
+      "title": "Connection Errors / s",
+      "transformations": [],
+      "type": "stat"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "prometheus"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "custom": {},
+          "mappings": [],
+          "unit": "bytes"
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 9,
+        "w": 12,
+        "x": 0,
+        "y": 7
+      },
+      "id": 5,
+      "options": {
+        "legend": {
+          "calcs": [],
+          "displayMode": "table",
+          "placement": "right"
+        },
+        "tooltip": {
+          "mode": "multi",
+          "sort": "desc"
+        }
+      },
+      "pluginVersion": "11.1.0",
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "prometheus"
+          },
+          "editorMode": "code",
+          "expr": "sum(go_memory_used_bytes)",
+          "legendFormat": "Used",
+          "refId": "A"
+        },
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "prometheus"
+          },
+          "editorMode": "code",
+          "expr": "go_memory_gc_goal_bytes",
+          "legendFormat": "GC Goal",
+          "refId": "B"
+        }
+      ],
+      "title": "Go Heap Usage vs GC Goal",
+      "type": "timeseries"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "prometheus"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "custom": {},
+          "decimals": 0,
+          "mappings": [],
+          "unit": "short"
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 9,
+        "w": 12,
+        "x": 12,
+        "y": 7
+      },
+      "id": 6,
+      "options": {
+        "legend": {
+          "calcs": [],
+          "displayMode": "table",
+          "placement": "right"
+        },
+        "tooltip": {
+          "mode": "multi",
+          "sort": "desc"
+        }
+      },
+      "pluginVersion": "11.1.0",
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "prometheus"
+          },
+          "editorMode": "code",
+          "expr": "rate(go_memory_allocations_total[$__rate_interval])",
+          "legendFormat": "Allocations/s",
+          "refId": "A"
+        },
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "prometheus"
+          },
+          "editorMode": "code",
+          "expr": "rate(go_memory_allocated_bytes_total[$__rate_interval])",
+          "legendFormat": "Allocated bytes/s",
+          "refId": "B"
+        }
+      ],
+      "title": "Allocation Activity",
+      "type": "timeseries"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "prometheus"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "custom": {},
+          "mappings": [],
+          "unit": "s"
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 9,
+        "w": 12,
+        "x": 0,
+        "y": 16
+      },
+      "id": 7,
+      "options": {
+        "legend": {
+          "calcs": [],
+          "displayMode": "table",
+          "placement": "right"
+        },
+        "tooltip": {
+          "mode": "multi",
+          "sort": "desc"
+        }
+      },
+      "pluginVersion": "11.1.0",
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "prometheus"
+          },
+          "editorMode": "code",
+          "expr": "histogram_quantile(0.5, sum(rate(http_server_request_duration_seconds_bucket[$__rate_interval])) by (le))",
+          "legendFormat": "p50",
+          "refId": "A"
+        },
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "prometheus"
+          },
+          "editorMode": "code",
+          "expr": "histogram_quantile(0.95, sum(rate(http_server_request_duration_seconds_bucket[$__rate_interval])) by (le))",
+          "legendFormat": "p95",
+          "refId": "B"
+        },
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "prometheus"
+          },
+          "editorMode": "code",
+          "expr": "histogram_quantile(0.99, sum(rate(http_server_request_duration_seconds_bucket[$__rate_interval])) by (le))",
+          "legendFormat": "p99",
+          "refId": "C"
+        }
+      ],
+      "title": "HTTP Request Duration Quantiles",
+      "type": "timeseries"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "prometheus"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "custom": {},
+          "mappings": [],
+          "unit": "ops"
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 9,
+        "w": 12,
+        "x": 12,
+        "y": 16
+      },
+      "id": 8,
+      "options": {
+        "legend": {
+          "calcs": [],
+          "displayMode": "table",
+          "placement": "right"
+        },
+        "tooltip": {
+          "mode": "multi",
+          "sort": "desc"
+        }
+      },
+      "pluginVersion": "11.1.0",
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "prometheus"
+          },
+          "editorMode": "code",
+          "expr": "sum(rate(http_server_request_duration_seconds_count[$__rate_interval])) by (http_response_status_code)",
+          "legendFormat": "{{http_response_status_code}}",
+          "refId": "A"
+        }
+      ],
+      "title": "HTTP Requests by Status",
+      "type": "timeseries"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "prometheus"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "custom": {},
+          "mappings": [],
+          "unit": "ops"
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 9,
+        "w": 12,
+        "x": 0,
+        "y": 25
+      },
+      "id": 9,
+      "options": {
+        "legend": {
+          "calcs": [],
+          "displayMode": "table",
+          "placement": "right"
+        },
+        "tooltip": {
+          "mode": "multi",
+          "sort": "desc"
+        }
+      },
+      "pluginVersion": "11.1.0",
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "prometheus"
+          },
+          "editorMode": "code",
+          "expr": "sum(rate(newt_connection_attempts_total{site_id=~\"$site_id\"}[$__rate_interval])) by (transport, result)",
+          "legendFormat": "{{transport}} • {{result}}",
+          "refId": "A"
+        }
+      ],
+      "title": "Connection Attempts by Transport/Result",
+      "type": "timeseries"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "prometheus"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "custom": {},
+          "mappings": [],
+          "unit": "ops"
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 9,
+        "w": 12,
+        "x": 12,
+        "y": 25
+      },
+      "id": 10,
+      "options": {
+        "legend": {
+          "calcs": [],
+          "displayMode": "table",
+          "placement": "right"
+        },
+        "tooltip": {
+          "mode": "multi",
+          "sort": "desc"
+        }
+      },
+      "pluginVersion": "11.1.0",
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "prometheus"
+          },
+          "editorMode": "code",
+          "expr": "sum(rate(newt_connection_errors_total{site_id=~\"$site_id\"}[$__rate_interval])) by (transport, error_type)",
+          "legendFormat": "{{transport}} • {{error_type}}",
+          "refId": "A"
+        }
+      ],
+      "title": "Connection Errors by Type",
+      "type": "timeseries"
+    },
+    {
+      "datasource": {
+        "type": "prometheus",
+        "uid": "prometheus"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "custom": {},
+          "decimals": 3,
+          "mappings": [],
+          "unit": "s"
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 9,
+        "w": 12,
+        "x": 0,
+        "y": 34
+      },
+      "id": 11,
+      "options": {
+        "legend": {
+          "calcs": [],
+          "displayMode": "table",
+          "placement": "right"
+        },
+        "tooltip": {
+          "mode": "multi",
+          "sort": "desc"
+        }
+      },
+      "pluginVersion": "11.1.0",
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "prometheus"
+          },
+          "editorMode": "code",
+          "expr": "histogram_quantile(0.5, sum(rate(newt_tunnel_latency_seconds_bucket{site_id=~\"$site_id\", tunnel_id=~\"$tunnel_id\"}[$__rate_interval])) by (le))",
+          "legendFormat": "p50",
+          "refId": "A"
+        },
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "prometheus"
+          },
+          "editorMode": "code",
+          "expr": "histogram_quantile(0.95, sum(rate(newt_tunnel_latency_seconds_bucket{site_id=~\"$site_id\", tunnel_id=~\"$tunnel_id\"}[$__rate_interval])) by (le))",
+          "legendFormat": "p95",
+          "refId": "B"
+        },
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "prometheus"
+          },
+          "editorMode": "code",
+          "expr": "histogram_quantile(0.99, sum(rate(newt_tunnel_latency_seconds_bucket{site_id=~\"$site_id\", tunnel_id=~\"$tunnel_id\"}[$__rate_interval])) by (le))",
+          "legendFormat": "p99",
+          "refId": "C"
+        }
+      ],
+      "title": "Tunnel Latency Quantiles",
+      "type": "timeseries"
+    },
+    {
+      "cards": {},
+      "color": {
+        "cardColor": "#b4ff00",
+        "colorScale": "sqrt",
+        "colorScheme": "interpolateTurbo"
+      },
+      "dataFormat": "tsbuckets",
+      "datasource": {
+        "type": "prometheus",
+        "uid": "prometheus"
+      },
+      "fieldConfig": {
+        "defaults": {
+          "custom": {},
+          "mappings": [],
+          "unit": "s"
+        },
+        "overrides": []
+      },
+      "gridPos": {
+        "h": 9,
+        "w": 12,
+        "x": 12,
+        "y": 34
+      },
+      "heatmap": {},
+      "hideZeroBuckets": true,
+      "id": 12,
+      "legend": {
+        "show": false
+      },
+      "options": {
+        "calculate": true,
+        "cellGap": 2,
+        "cellSize": "auto",
+        "color": {
+          "exponent": 0.5
+        },
+        "exemplars": {
+          "color": "rgba(255,255,255,0.7)"
+        },
+        "filterValues": {
+          "le": 1e-9
+        },
+        "legend": {
+          "show": false
+        },
+        "tooltip": {
+          "mode": "single",
+          "show": true
+        },
+        "xAxis": {
+          "show": true
+        },
+        "yAxis": {
+          "decimals": 3,
+          "show": true
+        }
+      },
+      "pluginVersion": "11.1.0",
+      "targets": [
+        {
+          "datasource": {
+            "type": "prometheus",
+            "uid": "prometheus"
+          },
+          "editorMode": "code",
+          "expr": "sum(rate(newt_tunnel_latency_seconds_bucket{site_id=~\"$site_id\", tunnel_id=~\"$tunnel_id\"}[$__rate_interval])) by (le)",
+          "format": "heatmap",
+          "legendFormat": "{{le}}",
+          "refId": "A"
+        }
+      ],
+      "title": "Tunnel Latency Bucket Rate",
+      "type": "heatmap"
+    }
+  ],
+  "refresh": "30s",
+  "schemaVersion": 39,
+  "style": "dark",
+  "tags": [
+    "newt",
+    "otel"
+  ],
+  "templating": {
+    "list": [
+      {
+        "current": {
+          "selected": false,
+          "text": "Prometheus",
+          "value": "prometheus"
+        },
+        "hide": 0,
+        "label": "Datasource",
+        "name": "DS_PROMETHEUS",
+        "options": [],
+        "query": "prometheus",
+        "refresh": 1,
+        "type": "datasource"
+      },
+      {
+        "current": {
+          "selected": false,
+          "text": "All",
+          "value": "$__all"
+        },
+        "datasource": {
+          "type": "prometheus",
+          "uid": "prometheus"
+        },
+        "definition": "label_values(target_info, site_id)",
+        "hide": 0,
+        "includeAll": true,
+        "label": "Site",
+        "multi": true,
+        "name": "site_id",
+        "options": [],
+        "query": {
+          "query": "label_values(target_info, site_id)",
+          "refId": "SiteIdVar"
+        },
+        "refresh": 2,
+        "regex": "",
+        "skipUrlSync": false,
+        "sort": 1,
+        "tagValuesQuery": "",
+        "tags": [],
+      "tagsQuery": "",
+        "type": "query",
+        "useTags": false
+      },
+      {
+        "current": {
+          "selected": false,
+          "text": "All",
+          "value": "$__all"
+        },
+        "datasource": {
+          "type": "prometheus",
+          "uid": "prometheus"
+        },
+        "definition": "label_values(newt_tunnel_latency_seconds_bucket{site_id=~\"$site_id\"}, tunnel_id)",
+        "hide": 0,
+        "includeAll": true,
+        "label": "Tunnel",
+        "multi": true,
+        "name": "tunnel_id",
+        "options": [],
+        "query": {
+          "query": "label_values(newt_tunnel_latency_seconds_bucket{site_id=~\"$site_id\"}, tunnel_id)",
+          "refId": "TunnelVar"
+        },
+        "refresh": 2,
+        "regex": "",
+        "skipUrlSync": false,
+        "sort": 1,
+        "tagValuesQuery": "",
+        "tags": [],
+        "tagsQuery": "",
+        "type": "query",
+        "useTags": false
+      }
+    ]
+  },
+  "time": {
+    "from": "now-6h",
+    "to": "now"
+  },
+  "timepicker": {
+    "refresh_intervals": [
+      "10s",
+      "30s",
+      "1m",
+      "5m",
+      "15m",
+      "30m",
+      "1h",
+      "2h",
+      "1d"
+    ],
+    "time_options": [
+      "5m",
+      "15m",
+      "1h",
+      "6h",
+      "12h",
+      "24h",
+      "2d",
+      "7d",
+      "30d"
+    ]
+  },
+  "timezone": "browser",
+  "title": "Newt Overview",
+  "uid": "newt-overview",
+  "version": 1,
+  "weekStart": ""
+}

examples/grafana/provisioning/dashboards/dashboard.yaml

@@ -0,0 +1,9 @@
+apiVersion: 1
+providers:
+  - name: "newt"
+    folder: "Newt"
+    type: file
+    disableDeletion: false
+    allowUiUpdates: true
+    options:
+      path: /var/lib/grafana/dashboards

examples/grafana/provisioning/datasources/prometheus.yaml

@@ -0,0 +1,9 @@
+apiVersion: 1
+datasources:
+  - name: Prometheus
+    type: prometheus
+    access: proxy
+    url: http://prometheus:9090
+    uid: prometheus
+    isDefault: true
+    editable: true

examples/logger_examples.go

@@ -0,0 +1,161 @@
+// Example usage patterns for the extensible logger
+package main
+
+import (
+	"fmt"
+	"os"
+	"time"
+
+	"github.com/fosrl/newt/logger"
+)
+
+// Example 1: Using the default logger (works exactly as before)
+func exampleDefaultLogger() {
+	logger.Info("Starting application")
+	logger.Debug("Debug information")
+	logger.Warn("Warning message")
+	logger.Error("Error occurred")
+}
+
+// Example 2: Using a custom logger instance with standard writer
+func exampleCustomInstance() {
+	log := logger.NewLogger()
+	log.SetLevel(logger.INFO)
+	log.Info("This is from a custom instance")
+}
+
+// Example 3: Custom writer that adds JSON formatting
+type JSONWriter struct{}
+
+func (w *JSONWriter) Write(level logger.LogLevel, timestamp time.Time, message string) {
+	fmt.Printf("{\"time\":\"%s\",\"level\":\"%s\",\"message\":\"%s\"}\n",
+		timestamp.Format(time.RFC3339),
+		level.String(),
+		message)
+}
+
+func exampleJSONLogger() {
+	jsonWriter := &JSONWriter{}
+	log := logger.NewLoggerWithWriter(jsonWriter)
+	log.Info("This will be logged as JSON")
+}
+
+// Example 4: File writer
+type FileWriter struct {
+	file *os.File
+}
+
+func NewFileWriter(filename string) (*FileWriter, error) {
+	file, err := os.OpenFile(filename, os.O_CREATE|os.O_WRONLY|os.O_APPEND, 0644)
+	if err != nil {
+		return nil, err
+	}
+	return &FileWriter{file: file}, nil
+}
+
+func (w *FileWriter) Write(level logger.LogLevel, timestamp time.Time, message string) {
+	fmt.Fprintf(w.file, "[%s] %s: %s\n",
+		timestamp.Format("2006-01-02 15:04:05"),
+		level.String(),
+		message)
+}
+
+func (w *FileWriter) Close() error {
+	return w.file.Close()
+}
+
+func exampleFileLogger() {
+	fileWriter, err := NewFileWriter("/tmp/app.log")
+	if err != nil {
+		panic(err)
+	}
+	defer fileWriter.Close()
+
+	log := logger.NewLoggerWithWriter(fileWriter)
+	log.Info("This goes to a file")
+}
+
+// Example 5: Multi-writer to log to multiple destinations
+type MultiWriter struct {
+	writers []logger.LogWriter
+}
+
+func NewMultiWriter(writers ...logger.LogWriter) *MultiWriter {
+	return &MultiWriter{writers: writers}
+}
+
+func (w *MultiWriter) Write(level logger.LogLevel, timestamp time.Time, message string) {
+	for _, writer := range w.writers {
+		writer.Write(level, timestamp, message)
+	}
+}
+
+func exampleMultiWriter() {
+	// Log to both stdout and a file
+	standardWriter := logger.NewStandardWriter()
+	fileWriter, _ := NewFileWriter("/tmp/app.log")
+
+	multiWriter := NewMultiWriter(standardWriter, fileWriter)
+	log := logger.NewLoggerWithWriter(multiWriter)
+
+	log.Info("This goes to both stdout and file!")
+}
+
+// Example 6: Conditional writer (only log errors to a specific destination)
+type ErrorOnlyWriter struct {
+	writer logger.LogWriter
+}
+
+func NewErrorOnlyWriter(writer logger.LogWriter) *ErrorOnlyWriter {
+	return &ErrorOnlyWriter{writer: writer}
+}
+
+func (w *ErrorOnlyWriter) Write(level logger.LogLevel, timestamp time.Time, message string) {
+	if level >= logger.ERROR {
+		w.writer.Write(level, timestamp, message)
+	}
+}
+
+func exampleConditionalWriter() {
+	errorWriter, _ := NewFileWriter("/tmp/errors.log")
+	errorOnlyWriter := NewErrorOnlyWriter(errorWriter)
+
+	log := logger.NewLoggerWithWriter(errorOnlyWriter)
+	log.Info("This won't be logged")
+	log.Error("This will be logged to errors.log")
+}
+
+/* Example 7: OS Log Writer (macOS/iOS only)
+// Uncomment on Darwin platforms
+
+func exampleOSLogWriter() {
+	osWriter := logger.NewOSLogWriter(
+		"net.pangolin.Pangolin.PacketTunnel",
+		"PangolinGo",
+		"MyApp",
+	)
+
+	log := logger.NewLoggerWithWriter(osWriter)
+	log.Info("This goes to os_log and can be viewed with Console.app")
+}
+*/
+
+func main() {
+	fmt.Println("=== Example 1: Default Logger ===")
+	exampleDefaultLogger()
+
+	fmt.Println("\n=== Example 2: Custom Instance ===")
+	exampleCustomInstance()
+
+	fmt.Println("\n=== Example 3: JSON Logger ===")
+	exampleJSONLogger()
+
+	fmt.Println("\n=== Example 4: File Logger ===")
+	exampleFileLogger()
+
+	fmt.Println("\n=== Example 5: Multi-Writer ===")
+	exampleMultiWriter()
+
+	fmt.Println("\n=== Example 6: Conditional Writer ===")
+	exampleConditionalWriter()
+}

examples/oslog_writer_example.go

@@ -0,0 +1,86 @@
+//go:build darwin
+// +build darwin
+
+package main
+
+/*
+#cgo CFLAGS: -I../PacketTunnel
+#include "../PacketTunnel/OSLogBridge.h"
+#include <stdlib.h>
+*/
+import "C"
+import (
+	"fmt"
+	"runtime"
+	"time"
+	"unsafe"
+)
+
+// OSLogWriter is a LogWriter implementation that writes to Apple's os_log
+type OSLogWriter struct {
+	subsystem string
+	category  string
+	prefix    string
+}
+
+// NewOSLogWriter creates a new OSLogWriter
+func NewOSLogWriter(subsystem, category, prefix string) *OSLogWriter {
+	writer := &OSLogWriter{
+		subsystem: subsystem,
+		category:  category,
+		prefix:    prefix,
+	}
+
+	// Initialize the OS log bridge
+	cSubsystem := C.CString(subsystem)
+	cCategory := C.CString(category)
+	defer C.free(unsafe.Pointer(cSubsystem))
+	defer C.free(unsafe.Pointer(cCategory))
+
+	C.initOSLogBridge(cSubsystem, cCategory)
+
+	return writer
+}
+
+// Write implements the LogWriter interface
+func (w *OSLogWriter) Write(level LogLevel, timestamp time.Time, message string) {
+	// Get caller information (skip 3 frames to get to the actual caller)
+	_, file, line, ok := runtime.Caller(3)
+	if !ok {
+		file = "unknown"
+		line = 0
+	} else {
+		// Get just the filename, not the full path
+		for i := len(file) - 1; i > 0; i-- {
+			if file[i] == '/' {
+				file = file[i+1:]
+				break
+			}
+		}
+	}
+
+	formattedTime := timestamp.Format("2006-01-02 15:04:05.000")
+	fullMessage := fmt.Sprintf("[%s] [%s] [%s] %s:%d - %s",
+		formattedTime, level.String(), w.prefix, file, line, message)
+
+	cMessage := C.CString(fullMessage)
+	defer C.free(unsafe.Pointer(cMessage))
+
+	// Map Go log levels to os_log levels:
+	// 0=DEBUG, 1=INFO, 2=DEFAULT (WARN), 3=ERROR
+	var osLogLevel C.int
+	switch level {
+	case DEBUG:
+		osLogLevel = 0 // DEBUG
+	case INFO:
+		osLogLevel = 1 // INFO
+	case WARN:
+		osLogLevel = 2 // DEFAULT
+	case ERROR, FATAL:
+		osLogLevel = 3 // ERROR
+	default:
+		osLogLevel = 2 // DEFAULT
+	}
+
+	C.logToOSLog(osLogLevel, cMessage)
+}

examples/otel-collector.yaml

@@ -0,0 +1,61 @@
+# Variant A: Direct scrape of Newt (/metrics) via Prometheus (no Collector needed)
+# Note: Newt already exposes labels like site_id, protocol, direction. Do not promote
+# resource attributes into labels when scraping Newt directly.
+#
+# Example Prometheus scrape config:
+# global:
+#   scrape_interval: 15s
+# scrape_configs:
+#   - job_name: newt
+#     static_configs:
+#       - targets: ["newt:2112"]
+#
+# Variant B: Use OTEL Collector (Newt -> OTLP -> Collector -> Prometheus)
+# This pipeline scrapes metrics from the Collector's Prometheus exporter.
+# Labels are already on datapoints; promotion from resource is OPTIONAL and typically NOT required.
+# If you enable transform/promote below, ensure you do not duplicate labels.
+
+receivers:
+  otlp:
+    protocols:
+      grpc:
+        endpoint: ":4317"
+
+processors:
+  memory_limiter:
+    check_interval: 5s
+    limit_percentage: 80
+    spike_limit_percentage: 25
+  resourcedetection:
+    detectors: [env, system]
+    timeout: 5s
+  batch: {}
+  # OPTIONAL: Only enable if you need to promote resource attributes to labels.
+  # WARNING: Newt already provides site_id as a label; avoid double-promotion.
+  # transform/promote:
+  #   error_mode: ignore
+  #   metric_statements:
+  #     - context: datapoint
+  #       statements:
+  #         - set(attributes["service_instance_id"], resource.attributes["service.instance.id"]) where resource.attributes["service.instance.id"] != nil
+  #         - set(attributes["site_id"], resource.attributes["site_id"]) where resource.attributes["site_id"] != nil
+
+exporters:
+  prometheus:
+    endpoint: ":8889"
+    send_timestamps: true
+  # prometheusremotewrite:
+  #   endpoint: http://mimir:9009/api/v1/push
+  debug:
+    verbosity: basic
+
+service:
+  pipelines:
+    metrics:
+      receivers: [otlp]
+      processors: [memory_limiter, resourcedetection, batch] # add transform/promote if you really need it
+      exporters: [prometheus]
+    traces:
+      receivers: [otlp]
+      processors: [memory_limiter, resourcedetection, batch]
+      exporters: [debug]

examples/prometheus.with-collector.yml

@@ -0,0 +1,16 @@
+global:
+  scrape_interval: 15s
+
+scrape_configs:
+  # IMPORTANT: Do not scrape Newt directly; scrape only the Collector!
+  - job_name: 'otel-collector'
+    static_configs:
+      - targets: ['otel-collector:8889']
+
+    # optional: limit metric cardinality
+    relabel_configs:
+      - action: labeldrop
+        regex: 'tunnel_id'
+      # - action: keep
+      #   source_labels: [site_id]
+      #   regex: '(site-a|site-b)'

examples/prometheus.yml

@@ -0,0 +1,21 @@
+global:
+  scrape_interval: 15s
+
+scrape_configs:
+  - job_name: 'newt'
+    scrape_interval: 15s
+    static_configs:
+      - targets: ['newt:2112']  # /metrics
+    relabel_configs:
+    # optional: drop tunnel_id
+      - action: labeldrop
+        regex: 'tunnel_id'
+    # optional: allow only specific sites
+      - action: keep
+        source_labels: [site_id]
+        regex: '(site-a|site-b)'
+
+  # WARNING: Do not enable this together with the 'newt' job above or you will double-count.
+  # - job_name: 'otel-collector'
+  #   static_configs:
+  #     - targets: ['otel-collector:8889']

flake.lock

@@ -0,0 +1,27 @@
+{
+  "nodes": {
+    "nixpkgs": {
+      "locked": {
+        "lastModified": 1763934636,
+        "narHash": "sha256-9glbI7f1uU+yzQCq5LwLgdZqx6svOhZWkd4JRY265fc=",
+        "owner": "NixOS",
+        "repo": "nixpkgs",
+        "rev": "ee09932cedcef15aaf476f9343d1dea2cb77e261",
+        "type": "github"
+      },
+      "original": {
+        "owner": "NixOS",
+        "ref": "nixpkgs-unstable",
+        "repo": "nixpkgs",
+        "type": "github"
+      }
+    },
+    "root": {
+      "inputs": {
+        "nixpkgs": "nixpkgs"
+      }
+    }
+  },
+  "root": "root",
+  "version": 7
+}

flake.nix

@@ -0,0 +1,102 @@
+{
+  description = "newt - A tunneling client for Pangolin";
+
+  inputs = {
+    nixpkgs.url = "github:NixOS/nixpkgs/nixpkgs-unstable";
+  };
+
+  outputs =
+    { self, nixpkgs }:
+    let
+      supportedSystems = [
+        "x86_64-linux"
+        "aarch64-linux"
+        "x86_64-darwin"
+        "aarch64-darwin"
+      ];
+      forAllSystems = nixpkgs.lib.genAttrs supportedSystems;
+      pkgsFor = system: nixpkgs.legacyPackages.${system};
+    in
+    {
+      packages = forAllSystems (
+        system:
+        let
+          pkgs = pkgsFor system;
+          inherit (pkgs) lib;
+
+          # Update version when releasing
+          version = "1.12.5";
+        in
+        {
+          default = self.packages.${system}.pangolin-newt;
+
+          pangolin-newt = pkgs.buildGoModule {
+            pname = "pangolin-newt";
+            inherit version;
+            src = pkgs.nix-gitignore.gitignoreSource [ ] ./.;
+
+            vendorHash = "sha256-X70emc3uPN2YpsbudtoeEZDHilaTvFMqfRaDmIgRVZE=";
+
+            nativeInstallCheckInputs = [ pkgs.versionCheckHook ];
+
+            env = {
+              CGO_ENABLED = 0;
+            };
+
+            ldflags = [
+              "-s"
+              "-w"
+              "-X main.newtVersion=${version}"
+            ];
+
+            # Tests are broken due to a lack of Internet.
+            # Disable running `go test`, and instead do
+            # a simple version check instead.
+            doCheck = false;
+            doInstallCheck = true;
+
+            versionCheckProgramArg = [ "-version" ];
+
+            meta = {
+              description = "A tunneling client for Pangolin";
+              homepage = "https://github.com/fosrl/newt";
+              license = lib.licenses.gpl3;
+              maintainers = [
+                lib.maintainers.water-sucks
+              ];
+              mainProgram = "newt";
+            };
+          };
+        }
+      );
+      devShells = forAllSystems (
+        system:
+        let
+          pkgs = pkgsFor system;
+
+          inherit (pkgs)
+            go
+            gopls
+            gotools
+            go-outline
+            gopkgs
+            godef
+            golint
+            ;
+        in
+        {
+          default = pkgs.mkShell {
+            buildInputs = [
+              go
+              gopls
+              gotools
+              go-outline
+              gopkgs
+              godef
+              golint
+            ];
+          };
+        }
+      );
+    };
+}

get-newt.sh

@@ -0,0 +1,334 @@
+#!/bin/sh
+
+# Get Newt - Cross-platform installation script
+# Usage: curl -fsSL https://raw.githubusercontent.com/fosrl/newt/refs/heads/main/get-newt.sh | sh
+
+set -e
+
+# Colors for output
+RED='\033[0;31m'
+GREEN='\033[0;32m'
+YELLOW='\033[1;33m'
+NC='\033[0m' # No Color
+
+# GitHub repository info
+REPO="fosrl/newt"
+GITHUB_API_URL="https://api.github.com/repos/${REPO}/releases/latest"
+
+# Function to print colored output
+print_status() {
+    printf '%b[INFO]%b %s\n' "${GREEN}" "${NC}" "$1"
+}
+
+print_warning() {
+    printf '%b[WARN]%b %s\n' "${YELLOW}" "${NC}" "$1"
+}
+
+print_error() {
+    printf '%b[ERROR]%b %s\n' "${RED}" "${NC}" "$1"
+}
+
+# Function to get latest version from GitHub API
+get_latest_version() {
+    latest_info=""
+
+    if command -v curl >/dev/null 2>&1; then
+        latest_info=$(curl -fsSL "$GITHUB_API_URL" 2>/dev/null)
+    elif command -v wget >/dev/null 2>&1; then
+        latest_info=$(wget -qO- "$GITHUB_API_URL" 2>/dev/null)
+    else
+        print_error "Neither curl nor wget is available."
+        exit 1
+    fi
+
+    if [ -z "$latest_info" ]; then
+        print_error "Failed to fetch latest version info"
+        exit 1
+    fi
+
+    version=$(printf '%s' "$latest_info" | grep '"tag_name"' | head -1 | sed 's/.*"tag_name": *"\([^"]*\)".*/\1/')
+
+    if [ -z "$version" ]; then
+        print_error "Could not parse version from GitHub API response"
+        exit 1
+    fi
+
+    version=$(printf '%s' "$version" | sed 's/^v//')
+    printf '%s' "$version"
+}
+
+# Detect OS and architecture
+detect_platform() {
+    os=""
+    arch=""
+
+    case "$(uname -s)" in
+        Linux*)     os="linux" ;;
+        Darwin*)    os="darwin" ;;
+        MINGW*|MSYS*|CYGWIN*) os="windows" ;;
+        FreeBSD*)   os="freebsd" ;;
+        *)
+            print_error "Unsupported operating system: $(uname -s)"
+            exit 1
+            ;;
+    esac
+
+    case "$(uname -m)" in
+        x86_64|amd64)   arch="amd64" ;;
+        arm64|aarch64)  arch="arm64" ;;
+        armv7l|armv6l)
+            if [ "$os" = "linux" ]; then
+                if [ "$(uname -m)" = "armv6l" ]; then
+                    arch="arm32v6"
+                else
+                    arch="arm32"
+                fi
+            else
+                arch="arm64"
+            fi
+            ;;
+        riscv64)
+            if [ "$os" = "linux" ]; then
+                arch="riscv64"
+            else
+                print_error "RISC-V architecture only supported on Linux"
+                exit 1
+            fi
+            ;;
+        *)
+            print_error "Unsupported architecture: $(uname -m)"
+            exit 1
+            ;;
+    esac
+
+    printf '%s_%s' "$os" "$arch"
+}
+
+# Determine installation directory (default fallback)
+get_install_dir() {
+    case "$PLATFORM" in
+        *windows*)
+            echo "$HOME/bin"
+            ;;
+        *)
+            echo "/usr/local/bin"
+            ;;
+    esac
+}
+
+# Parse --path argument from args
+# Returns the value after --path, or empty string if not provided
+parse_path_arg() {
+    while [ $# -gt 0 ]; do
+        case "$1" in
+            --path)
+                if [ -n "$2" ]; then
+                    printf '%s' "$2"
+                    return
+                fi
+                ;;
+            --path=*)
+                printf '%s' "${1#--path=}"
+                return
+                ;;
+        esac
+        shift
+    done
+}
+
+# Detect an existing newt binary location.
+# Tries unprivileged which first, then sudo which (for binaries only visible to root).
+# Returns the full path of the binary, or empty string if not found.
+detect_existing_binary() {
+    existing=""
+
+    # Try unprivileged which first
+    existing=$(command -v newt 2>/dev/null || true)
+    if [ -n "$existing" ]; then
+        printf '%s' "$existing"
+        return
+    fi
+
+    # Try sudo which — some installations land in paths only root can see in $PATH
+    if command -v sudo >/dev/null 2>&1; then
+        existing=$(sudo which newt 2>/dev/null || true)
+        if [ -n "$existing" ]; then
+            printf '%s' "$existing"
+            return
+        fi
+    fi
+}
+
+# Check if we need sudo for installation
+needs_sudo() {
+    install_dir="$1"
+    if [ -w "$install_dir" ] 2>/dev/null; then
+        return 1  # No sudo needed
+    else
+        return 0  # Sudo needed
+    fi
+}
+
+# Get the appropriate command prefix (sudo or empty)
+get_sudo_cmd() {
+    install_dir="$1"
+    if needs_sudo "$install_dir"; then
+        if command -v sudo >/dev/null 2>&1; then
+            echo "sudo"
+        else
+            print_error "Cannot write to ${install_dir} and sudo is not available."
+            print_error "Please run this script as root or install sudo."
+            exit 1
+        fi
+    else
+        echo ""
+    fi
+}
+
+# Download and install newt
+install_newt() {
+    platform="$1"
+    install_dir="$2"
+    sudo_cmd="$3"
+    custom_path="$4"
+    binary_name="newt_${platform}"
+    final_name="newt"
+
+    case "$platform" in
+        *windows*)
+            binary_name="${binary_name}.exe"
+            final_name="newt.exe"
+            ;;
+    esac
+
+    download_url="${BASE_URL}/${binary_name}"
+    temp_file="/tmp/${final_name}"
+
+    # If a custom path is provided, use it directly; otherwise use install_dir/final_name
+    if [ -n "$custom_path" ]; then
+        final_path="$custom_path"
+        install_dir=$(dirname "$final_path")
+    else
+        final_path="${install_dir}/${final_name}"
+    fi
+
+    print_status "Downloading newt from ${download_url}"
+
+    if command -v curl >/dev/null 2>&1; then
+        curl -fsSL "$download_url" -o "$temp_file"
+    elif command -v wget >/dev/null 2>&1; then
+        wget -q "$download_url" -O "$temp_file"
+    else
+        print_error "Neither curl nor wget is available."
+        exit 1
+    fi
+
+    # Make executable before moving
+    chmod +x "$temp_file"
+
+    # Create install directory if it doesn't exist and move binary
+    if [ -n "$sudo_cmd" ]; then
+        $sudo_cmd mkdir -p "$install_dir"
+        print_status "Using sudo to install to ${install_dir}"
+        $sudo_cmd mv "$temp_file" "$final_path"
+    else
+        mkdir -p "$install_dir"
+        mv "$temp_file" "$final_path"
+    fi
+
+    print_status "newt installed to ${final_path}"
+
+    # Check if install directory is in PATH
+    if ! echo "$PATH" | grep -q "$install_dir"; then
+        print_warning "Install directory ${install_dir} is not in your PATH."
+        print_warning "Add it with:"
+        print_warning "  export PATH=\"${install_dir}:\$PATH\""
+    fi
+}
+
+# Verify installation
+verify_installation() {
+    install_dir="$1"
+    exe_suffix=""
+
+    case "$PLATFORM" in
+        *windows*) exe_suffix=".exe" ;;
+    esac
+
+    newt_path="${install_dir}/newt${exe_suffix}"
+
+    if [ -x "$newt_path" ]; then
+        print_status "Installation successful!"
+        print_status "newt version: $("$newt_path" --version 2>/dev/null || printf 'unknown')"
+        return 0
+    else
+        print_error "Installation failed. Binary not found or not executable."
+        return 1
+    fi
+}
+
+# Main installation process
+main() {
+    # --path explicitly overrides everything
+    CUSTOM_PATH=$(parse_path_arg "$@")
+
+    if [ -n "$CUSTOM_PATH" ]; then
+        print_status "Installing latest version of newt to ${CUSTOM_PATH}..."
+    else
+        print_status "Installing latest version of newt..."
+    fi
+
+    print_status "Fetching latest version..."
+    VERSION=$(get_latest_version)
+    print_status "Latest version: v${VERSION}"
+
+    BASE_URL="https://github.com/${REPO}/releases/download/${VERSION}"
+
+    PLATFORM=$(detect_platform)
+    print_status "Detected platform: ${PLATFORM}"
+
+    if [ -n "$CUSTOM_PATH" ]; then
+        # --path wins; derive INSTALL_DIR from it
+        INSTALL_DIR=$(dirname "$CUSTOM_PATH")
+    else
+        # Try to find an existing installation so we update the right place
+        EXISTING_BINARY=$(detect_existing_binary)
+        if [ -n "$EXISTING_BINARY" ]; then
+            print_status "Found existing newt binary at ${EXISTING_BINARY}"
+            CUSTOM_PATH="$EXISTING_BINARY"
+            INSTALL_DIR=$(dirname "$EXISTING_BINARY")
+            print_status "Will update existing installation at ${INSTALL_DIR}"
+        else
+            INSTALL_DIR=$(get_install_dir)
+        fi
+    fi
+
+    print_status "Install directory: ${INSTALL_DIR}"
+
+    # Check if we need sudo
+    SUDO_CMD=$(get_sudo_cmd "$INSTALL_DIR")
+    if [ -n "$SUDO_CMD" ]; then
+        print_status "Root privileges required for installation to ${INSTALL_DIR}"
+    fi
+
+    install_newt "$PLATFORM" "$INSTALL_DIR" "$SUDO_CMD" "$CUSTOM_PATH"
+
+    if [ -n "$CUSTOM_PATH" ]; then
+        if [ -x "$CUSTOM_PATH" ]; then
+            print_status "Installation successful!"
+            print_status "newt version: $("$CUSTOM_PATH" --version 2>/dev/null || printf 'unknown')"
+            print_status "newt is ready to use!"
+        else
+            print_error "Installation failed. Binary not found or not executable at ${CUSTOM_PATH}."
+            exit 1
+        fi
+    elif verify_installation "$INSTALL_DIR"; then
+        print_status "newt is ready to use!"
+        print_status "Run 'newt --help' to get started."
+    else
+        exit 1
+    fi
+}
+
+# Run main function
+main "$@"

go.mod

@@ -1,22 +1,75 @@
 module github.com/fosrl/newt
 
-go 1.23.1
-
-toolchain go1.23.2
+go 1.25.0
 
 require (
+	github.com/coder/websocket v1.8.14
+	github.com/creack/pty v1.1.24
+	github.com/gaissmai/bart v0.26.1
+	github.com/go-crypt/crypt v0.14.15
+	github.com/go-crypt/x v0.4.16
 	github.com/gorilla/websocket v1.5.3
-	golang.org/x/net v0.30.0
-	golang.zx2c4.com/wireguard v0.0.0-20231211153847-12269c276173
-	golang.zx2c4.com/wireguard/wgctrl v0.0.0-20230429144221-925a1e7659e6
-	gvisor.dev/gvisor v0.0.0-20230927004350-cbd86285d259
+	github.com/moby/moby/api v1.54.2
+	github.com/moby/moby/client v0.4.1
+	github.com/prometheus/client_golang v1.23.2
+	github.com/vishvananda/netlink v1.3.1
+	go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.68.0
+	go.opentelemetry.io/contrib/instrumentation/runtime v0.68.0
+	go.opentelemetry.io/otel v1.43.0
+	go.opentelemetry.io/otel/exporters/otlp/otlpmetric/otlpmetricgrpc v1.43.0
+	go.opentelemetry.io/otel/exporters/otlp/otlptrace/otlptracegrpc v1.43.0
+	go.opentelemetry.io/otel/exporters/prometheus v0.65.0
+	go.opentelemetry.io/otel/metric v1.43.0
+	go.opentelemetry.io/otel/sdk v1.43.0
+	go.opentelemetry.io/otel/sdk/metric v1.43.0
+	golang.org/x/crypto v0.52.0
+	golang.org/x/exp v0.0.0-20251113190631-e25ba8c21ef6
+	golang.org/x/net v0.55.0
+	golang.org/x/sys v0.45.0
+	golang.zx2c4.com/wireguard v0.0.0-20250521234502-f333402bd9cb
+	golang.zx2c4.com/wireguard/wgctrl v0.0.0-20241231184526-a9ab2273dd10
+	golang.zx2c4.com/wireguard/windows v1.0.1
+	google.golang.org/grpc v1.81.1
+	gopkg.in/yaml.v3 v3.0.1
+	gvisor.dev/gvisor v0.0.0-20250503011706-39ed1f5ac29c
+	software.sslmate.com/src/go-pkcs12 v0.7.1
 )
 
 require (
-	github.com/google/btree v1.1.2 // indirect
-	github.com/google/go-cmp v0.6.0 // indirect
-	golang.org/x/crypto v0.28.0 // indirect
-	golang.org/x/sys v0.26.0 // indirect
-	golang.org/x/time v0.7.0 // indirect
+	github.com/Microsoft/go-winio v0.6.2 // indirect
+	github.com/beorn7/perks v1.0.1 // indirect
+	github.com/cenkalti/backoff/v5 v5.0.3 // indirect
+	github.com/cespare/xxhash/v2 v2.3.0 // indirect
+	github.com/containerd/errdefs v1.0.0 // indirect
+	github.com/containerd/errdefs/pkg v0.3.0 // indirect
+	github.com/distribution/reference v0.6.0 // indirect
+	github.com/docker/go-connections v0.7.0 // indirect
+	github.com/docker/go-units v0.5.0 // indirect
+	github.com/felixge/httpsnoop v1.0.4 // indirect
+	github.com/fsnotify/fsnotify v1.9.0 // indirect
+	github.com/go-logr/logr v1.4.3 // indirect
+	github.com/go-logr/stdr v1.2.2 // indirect
+	github.com/google/btree v1.1.3 // indirect
+	github.com/google/uuid v1.6.0 // indirect
+	github.com/grpc-ecosystem/grpc-gateway/v2 v2.28.0 // indirect
+	github.com/moby/docker-image-spec v1.3.1 // indirect
+	github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822 // indirect
+	github.com/opencontainers/go-digest v1.0.0 // indirect
+	github.com/opencontainers/image-spec v1.1.1 // indirect
+	github.com/prometheus/client_model v0.6.2 // indirect
+	github.com/prometheus/common v0.67.5 // indirect
+	github.com/prometheus/otlptranslator v1.0.0 // indirect
+	github.com/prometheus/procfs v0.20.1 // indirect
+	github.com/vishvananda/netns v0.0.5 // indirect
+	go.opentelemetry.io/auto/sdk v1.2.1 // indirect
+	go.opentelemetry.io/otel/exporters/otlp/otlptrace v1.43.0 // indirect
+	go.opentelemetry.io/otel/trace v1.43.0 // indirect
+	go.opentelemetry.io/proto/otlp v1.10.0 // indirect
+	go.yaml.in/yaml/v2 v2.4.4 // indirect
+	golang.org/x/text v0.37.0 // indirect
+	golang.org/x/time v0.12.0 // indirect
 	golang.zx2c4.com/wintun v0.0.0-20230126152724-0fa3db229ce2 // indirect
+	google.golang.org/genproto/googleapis/api v0.0.0-20260401024825-9d38bb4040a9 // indirect
+	google.golang.org/genproto/googleapis/rpc v0.0.0-20260401024825-9d38bb4040a9 // indirect
+	google.golang.org/protobuf v1.36.11 // indirect
 )

go.sum

@@ -1,22 +1,168 @@
-github.com/google/btree v1.1.2 h1:xf4v41cLI2Z6FxbKm+8Bu+m8ifhj15JuZ9sa0jZCMUU=
-github.com/google/btree v1.1.2/go.mod h1:qOPhT0dTNdNzV6Z/lhRX0YXUafgPLFUh+gZMl761Gm4=
-github.com/google/go-cmp v0.6.0 h1:ofyhxvXcZhMsU5ulbFiLKl/XBFqE1GSq7atu8tAmTRI=
-github.com/google/go-cmp v0.6.0/go.mod h1:17dUlkBOakJ0+DkrSSNjCkIjxS6bF9zb3elmeNGIjoY=
+github.com/Microsoft/go-winio v0.6.2 h1:F2VQgta7ecxGYO8k3ZZz3RS8fVIXVxONVUPlNERoyfY=
+github.com/Microsoft/go-winio v0.6.2/go.mod h1:yd8OoFMLzJbo9gZq8j5qaps8bJ9aShtEA8Ipt1oGCvU=
+github.com/beorn7/perks v1.0.1 h1:VlbKKnNfV8bJzeqoa4cOKqO6bYr3WgKZxO8Z16+hsOM=
+github.com/beorn7/perks v1.0.1/go.mod h1:G2ZrVWU2WbWT9wwq4/hrbKbnv/1ERSJQ0ibhJ6rlkpw=
+github.com/cenkalti/backoff/v5 v5.0.3 h1:ZN+IMa753KfX5hd8vVaMixjnqRZ3y8CuJKRKj1xcsSM=
+github.com/cenkalti/backoff/v5 v5.0.3/go.mod h1:rkhZdG3JZukswDf7f0cwqPNk4K0sa+F97BxZthm/crw=
+github.com/cespare/xxhash/v2 v2.3.0 h1:UL815xU9SqsFlibzuggzjXhog7bL6oX9BbNZnL2UFvs=
+github.com/cespare/xxhash/v2 v2.3.0/go.mod h1:VGX0DQ3Q6kWi7AoAeZDth3/j3BFtOZR5XLFGgcrjCOs=
+github.com/coder/websocket v1.8.14 h1:9L0p0iKiNOibykf283eHkKUHHrpG7f65OE3BhhO7v9g=
+github.com/coder/websocket v1.8.14/go.mod h1:NX3SzP+inril6yawo5CQXx8+fk145lPDC6pumgx0mVg=
+github.com/containerd/errdefs v1.0.0 h1:tg5yIfIlQIrxYtu9ajqY42W3lpS19XqdxRQeEwYG8PI=
+github.com/containerd/errdefs v1.0.0/go.mod h1:+YBYIdtsnF4Iw6nWZhJcqGSg/dwvV7tyJ/kCkyJ2k+M=
+github.com/containerd/errdefs/pkg v0.3.0 h1:9IKJ06FvyNlexW690DXuQNx2KA2cUJXx151Xdx3ZPPE=
+github.com/containerd/errdefs/pkg v0.3.0/go.mod h1:NJw6s9HwNuRhnjJhM7pylWwMyAkmCQvQ4GpJHEqRLVk=
+github.com/creack/pty v1.1.24 h1:bJrF4RRfyJnbTJqzRLHzcGaZK1NeM5kTC9jGgovnR1s=
+github.com/creack/pty v1.1.24/go.mod h1:08sCNb52WyoAwi2QDyzUCTgcvVFhUzewun7wtTfvcwE=
+github.com/davecgh/go-spew v1.1.1 h1:vj9j/u1bqnvCEfJOwUhtlOARqs3+rkHYY13jYWTU97c=
+github.com/davecgh/go-spew v1.1.1/go.mod h1:J7Y8YcW2NihsgmVo/mv3lAwl/skON4iLHjSsI+c5H38=
+github.com/distribution/reference v0.6.0 h1:0IXCQ5g4/QMHHkarYzh5l+u8T3t73zM5QvfrDyIgxBk=
+github.com/distribution/reference v0.6.0/go.mod h1:BbU0aIcezP1/5jX/8MP0YiH4SdvB5Y4f/wlDRiLyi3E=
+github.com/docker/go-connections v0.7.0 h1:6SsRfJddP22WMrCkj19x9WKjEDTB+ahsdiGYf0mN39c=
+github.com/docker/go-connections v0.7.0/go.mod h1:no1qkHdjq7kLMGUXYAduOhYPSJxxvgWBh7ogVvptn3Q=
+github.com/docker/go-units v0.5.0 h1:69rxXcBk27SvSaaxTtLh/8llcHD8vYHT7WSdRZ/jvr4=
+github.com/docker/go-units v0.5.0/go.mod h1:fgPhTUdO+D/Jk86RDLlptpiXQzgHJF7gydDDbaIK4Dk=
+github.com/felixge/httpsnoop v1.0.4 h1:NFTV2Zj1bL4mc9sqWACXbQFVBBg2W3GPvqp8/ESS2Wg=
+github.com/felixge/httpsnoop v1.0.4/go.mod h1:m8KPJKqk1gH5J9DgRY2ASl2lWCfGKXixSwevea8zH2U=
+github.com/fsnotify/fsnotify v1.9.0 h1:2Ml+OJNzbYCTzsxtv8vKSFD9PbJjmhYF14k/jKC7S9k=
+github.com/fsnotify/fsnotify v1.9.0/go.mod h1:8jBTzvmWwFyi3Pb8djgCCO5IBqzKJ/Jwo8TRcHyHii0=
+github.com/gaissmai/bart v0.26.1 h1:+w4rnLGNlA2GDVn382Tfe3jOsK5vOr5n4KmigJ9lbTo=
+github.com/gaissmai/bart v0.26.1/go.mod h1:GREWQfTLRWz/c5FTOsIw+KkscuFkIV5t8Rp7Nd1Td5c=
+github.com/go-crypt/crypt v0.14.15 h1:q1i5OMpL05r935IxWmXgpDAVF0nvi4SMoHhGXLBQUEQ=
+github.com/go-crypt/crypt v0.14.15/go.mod h1:0n/to1VqIZPENj2yEUa/sLLYYnmupma6cp+QMX4zfF0=
+github.com/go-crypt/x v0.4.16 h1:WXdY28H/0MsXnH+gwerxuCcvBTJPkBG90u6oS4gIPZI=
+github.com/go-crypt/x v0.4.16/go.mod h1:vmVFA/d/oLrEaCbqsLcjBMlTqF8u8pvH/c4+EJ/ped8=
+github.com/go-logr/logr v1.2.2/go.mod h1:jdQByPbusPIv2/zmleS9BjJVeZ6kBagPoEUsqbVz/1A=
+github.com/go-logr/logr v1.4.3 h1:CjnDlHq8ikf6E492q6eKboGOC0T8CDaOvkHCIg8idEI=
+github.com/go-logr/logr v1.4.3/go.mod h1:9T104GzyrTigFIr8wt5mBrctHMim0Nb2HLGrmQ40KvY=
+github.com/go-logr/stdr v1.2.2 h1:hSWxHoqTgW2S2qGc0LTAI563KZ5YKYRhT3MFKZMbjag=
+github.com/go-logr/stdr v1.2.2/go.mod h1:mMo/vtBO5dYbehREoey6XUKy/eSumjCCveDpRre4VKE=
+github.com/golang/protobuf v1.5.4 h1:i7eJL8qZTpSEXOPTxNKhASYpMn+8e5Q6AdndVa1dWek=
+github.com/golang/protobuf v1.5.4/go.mod h1:lnTiLA8Wa4RWRcIUkrtSVa5nRhsEGBg48fD6rSs7xps=
+github.com/google/btree v1.1.3 h1:CVpQJjYgC4VbzxeGVHfvZrv1ctoYCAI8vbl07Fcxlyg=
+github.com/google/btree v1.1.3/go.mod h1:qOPhT0dTNdNzV6Z/lhRX0YXUafgPLFUh+gZMl761Gm4=
+github.com/google/go-cmp v0.7.0 h1:wk8382ETsv4JYUZwIsn6YpYiWiBsYLSJiTsyBybVuN8=
+github.com/google/go-cmp v0.7.0/go.mod h1:pXiqmnSA92OHEEa9HXL2W4E7lf9JzCmGVUdgjX3N/iU=
+github.com/google/uuid v1.6.0 h1:NIvaJDMOsjHA8n1jAhLSgzrAzy1Hgr+hNrb57e+94F0=
+github.com/google/uuid v1.6.0/go.mod h1:TIyPZe4MgqvfeYDBFedMoGGpEw/LqOeaOT+nhxU+yHo=
 github.com/gorilla/websocket v1.5.3 h1:saDtZ6Pbx/0u+bgYQ3q96pZgCzfhKXGPqt7kZ72aNNg=
 github.com/gorilla/websocket v1.5.3/go.mod h1:YR8l580nyteQvAITg2hZ9XVh4b55+EU/adAjf1fMHhE=
-golang.org/x/crypto v0.28.0 h1:GBDwsMXVQi34v5CCYUm2jkJvu4cbtru2U4TN2PSyQnw=
-golang.org/x/crypto v0.28.0/go.mod h1:rmgy+3RHxRZMyY0jjAJShp2zgEdOqj2AO7U0pYmeQ7U=
-golang.org/x/net v0.30.0 h1:AcW1SDZMkb8IpzCdQUaIq2sP4sZ4zw+55h6ynffypl4=
-golang.org/x/net v0.30.0/go.mod h1:2wGyMJ5iFasEhkwi13ChkO/t1ECNC4X4eBKkVFyYFlU=
-golang.org/x/sys v0.26.0 h1:KHjCJyddX0LoSTb3J+vWpupP9p0oznkqVk/IfjymZbo=
-golang.org/x/sys v0.26.0/go.mod h1:/VUhepiaJMQUp4+oa/7Zr1D23ma6VTLIYjOOTFZPUcA=
-golang.org/x/time v0.7.0 h1:ntUhktv3OPE6TgYxXWv9vKvUSJyIFJlyohwbkEwPrKQ=
-golang.org/x/time v0.7.0/go.mod h1:3BpzKBy/shNhVucY/MWOyx10tF3SFh9QdLuxbVysPQM=
+github.com/grpc-ecosystem/grpc-gateway/v2 v2.28.0 h1:HWRh5R2+9EifMyIHV7ZV+MIZqgz+PMpZ14Jynv3O2Zs=
+github.com/grpc-ecosystem/grpc-gateway/v2 v2.28.0/go.mod h1:JfhWUomR1baixubs02l85lZYYOm7LV6om4ceouMv45c=
+github.com/klauspost/compress v1.18.0 h1:c/Cqfb0r+Yi+JtIEq73FWXVkRonBlf0CRNYc8Zttxdo=
+github.com/klauspost/compress v1.18.0/go.mod h1:2Pp+KzxcywXVXMr50+X0Q/Lsb43OQHYWRCY2AiWywWQ=
+github.com/kr/pretty v0.3.1 h1:flRD4NNwYAUpkphVc1HcthR4KEIFJ65n8Mw5qdRn3LE=
+github.com/kr/pretty v0.3.1/go.mod h1:hoEshYVHaxMs3cyo3Yncou5ZscifuDolrwPKZanG3xk=
+github.com/kr/text v0.2.0 h1:5Nx0Ya0ZqY2ygV366QzturHI13Jq95ApcVaJBhpS+AY=
+github.com/kr/text v0.2.0/go.mod h1:eLer722TekiGuMkidMxC/pM04lWEeraHUUmBw8l2grE=
+github.com/kylelemons/godebug v1.1.0 h1:RPNrshWIDI6G2gRW9EHilWtl7Z6Sb1BR0xunSBf0SNc=
+github.com/kylelemons/godebug v1.1.0/go.mod h1:9/0rRGxNHcop5bhtWyNeEfOS8JIWk580+fNqagV/RAw=
+github.com/moby/docker-image-spec v1.3.1 h1:jMKff3w6PgbfSa69GfNg+zN/XLhfXJGnEx3Nl2EsFP0=
+github.com/moby/docker-image-spec v1.3.1/go.mod h1:eKmb5VW8vQEh/BAr2yvVNvuiJuY6UIocYsFu/DxxRpo=
+github.com/moby/moby/api v1.54.2 h1:wiat9QAhnDQjA7wk1kh/TqHz2I1uUA7M7t9SAl/JNXg=
+github.com/moby/moby/api v1.54.2/go.mod h1:+RQ6wluLwtYaTd1WnPLykIDPekkuyD/ROWQClE83pzs=
+github.com/moby/moby/client v0.4.1 h1:DMQgisVoMkmMs7fp3ROSdiBnoAu8+vo3GggFl06M/wY=
+github.com/moby/moby/client v0.4.1/go.mod h1:z52C9O2POPOsnxZAy//WtKcQ32P+jT/NGeXu/7nfjGQ=
+github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822 h1:C3w9PqII01/Oq1c1nUAm88MOHcQC9l5mIlSMApZMrHA=
+github.com/munnerz/goautoneg v0.0.0-20191010083416-a7dc8b61c822/go.mod h1:+n7T8mK8HuQTcFwEeznm/DIxMOiR9yIdICNftLE1DvQ=
+github.com/opencontainers/go-digest v1.0.0 h1:apOUWs51W5PlhuyGyz9FCeeBIOUDA/6nW8Oi/yOhh5U=
+github.com/opencontainers/go-digest v1.0.0/go.mod h1:0JzlMkj0TRzQZfJkVvzbP0HBR3IKzErnv2BNG4W4MAM=
+github.com/opencontainers/image-spec v1.1.1 h1:y0fUlFfIZhPF1W537XOLg0/fcx6zcHCJwooC2xJA040=
+github.com/opencontainers/image-spec v1.1.1/go.mod h1:qpqAh3Dmcf36wStyyWU+kCeDgrGnAve2nCC8+7h8Q0M=
+github.com/pmezard/go-difflib v1.0.0 h1:4DBwDE0NGyQoBHbLQYPwSUPoCMWR5BEzIk/f1lZbAQM=
+github.com/pmezard/go-difflib v1.0.0/go.mod h1:iKH77koFhYxTK1pcRnkKkqfTogsbg7gZNVY4sRDYZ/4=
+github.com/prometheus/client_golang v1.23.2 h1:Je96obch5RDVy3FDMndoUsjAhG5Edi49h0RJWRi/o0o=
+github.com/prometheus/client_golang v1.23.2/go.mod h1:Tb1a6LWHB3/SPIzCoaDXI4I8UHKeFTEQ1YCr+0Gyqmg=
+github.com/prometheus/client_model v0.6.2 h1:oBsgwpGs7iVziMvrGhE53c/GrLUsZdHnqNwqPLxwZyk=
+github.com/prometheus/client_model v0.6.2/go.mod h1:y3m2F6Gdpfy6Ut/GBsUqTWZqCUvMVzSfMLjcu6wAwpE=
+github.com/prometheus/common v0.67.5 h1:pIgK94WWlQt1WLwAC5j2ynLaBRDiinoAb86HZHTUGI4=
+github.com/prometheus/common v0.67.5/go.mod h1:SjE/0MzDEEAyrdr5Gqc6G+sXI67maCxzaT3A2+HqjUw=
+github.com/prometheus/otlptranslator v1.0.0 h1:s0LJW/iN9dkIH+EnhiD3BlkkP5QVIUVEoIwkU+A6qos=
+github.com/prometheus/otlptranslator v1.0.0/go.mod h1:vRYWnXvI6aWGpsdY/mOT/cbeVRBlPWtBNDb7kGR3uKM=
+github.com/prometheus/procfs v0.20.1 h1:XwbrGOIplXW/AU3YhIhLODXMJYyC1isLFfYCsTEycfc=
+github.com/prometheus/procfs v0.20.1/go.mod h1:o9EMBZGRyvDrSPH1RqdxhojkuXstoe4UlK79eF5TGGo=
+github.com/rogpeppe/go-internal v1.14.1 h1:UQB4HGPB6osV0SQTLymcB4TgvyWu6ZyliaW0tI/otEQ=
+github.com/rogpeppe/go-internal v1.14.1/go.mod h1:MaRKkUm5W0goXpeCfT7UZI6fk/L7L7so1lCWt35ZSgc=
+github.com/stretchr/testify v1.11.1 h1:7s2iGBzp5EwR7/aIZr8ao5+dra3wiQyKjjFuvgVKu7U=
+github.com/stretchr/testify v1.11.1/go.mod h1:wZwfW3scLgRK+23gO65QZefKpKQRnfz6sD981Nm4B6U=
+github.com/vishvananda/netlink v1.3.1 h1:3AEMt62VKqz90r0tmNhog0r/PpWKmrEShJU0wJW6bV0=
+github.com/vishvananda/netlink v1.3.1/go.mod h1:ARtKouGSTGchR8aMwmkzC0qiNPrrWO5JS/XMVl45+b4=
+github.com/vishvananda/netns v0.0.5 h1:DfiHV+j8bA32MFM7bfEunvT8IAqQ/NzSJHtcmW5zdEY=
+github.com/vishvananda/netns v0.0.5/go.mod h1:SpkAiCQRtJ6TvvxPnOSyH3BMl6unz3xZlaprSwhNNJM=
+go.opentelemetry.io/auto/sdk v1.2.1 h1:jXsnJ4Lmnqd11kwkBV2LgLoFMZKizbCi5fNZ/ipaZ64=
+go.opentelemetry.io/auto/sdk v1.2.1/go.mod h1:KRTj+aOaElaLi+wW1kO/DZRXwkF4C5xPbEe3ZiIhN7Y=
+go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.68.0 h1:CqXxU8VOmDefoh0+ztfGaymYbhdB/tT3zs79QaZTNGY=
+go.opentelemetry.io/contrib/instrumentation/net/http/otelhttp v0.68.0/go.mod h1:BuhAPThV8PBHBvg8ZzZ/Ok3idOdhWIodywz2xEcRbJo=
+go.opentelemetry.io/contrib/instrumentation/runtime v0.68.0 h1:jhVIQEprwUTV+KfzzliLidclhoTOoHTgdz96kAyR8mU=
+go.opentelemetry.io/contrib/instrumentation/runtime v0.68.0/go.mod h1:4HsdbLUbernaTnA8CNaNE+1g026SciXb3juRYe3l8EY=
+go.opentelemetry.io/otel v1.43.0 h1:mYIM03dnh5zfN7HautFE4ieIig9amkNANT+xcVxAj9I=
+go.opentelemetry.io/otel v1.43.0/go.mod h1:JuG+u74mvjvcm8vj8pI5XiHy1zDeoCS2LB1spIq7Ay0=
+go.opentelemetry.io/otel/exporters/otlp/otlpmetric/otlpmetricgrpc v1.43.0 h1:8UQVDcZxOJLtX6gxtDt3vY2WTgvZqMQRzjsqiIHQdkc=
+go.opentelemetry.io/otel/exporters/otlp/otlpmetric/otlpmetricgrpc v1.43.0/go.mod h1:2lmweYCiHYpEjQ/lSJBYhj9jP1zvCvQW4BqL9dnT7FQ=
+go.opentelemetry.io/otel/exporters/otlp/otlptrace v1.43.0 h1:88Y4s2C8oTui1LGM6bTWkw0ICGcOLCAI5l6zsD1j20k=
+go.opentelemetry.io/otel/exporters/otlp/otlptrace v1.43.0/go.mod h1:Vl1/iaggsuRlrHf/hfPJPvVag77kKyvrLeD10kpMl+A=
+go.opentelemetry.io/otel/exporters/otlp/otlptrace/otlptracegrpc v1.43.0 h1:RAE+JPfvEmvy+0LzyUA25/SGawPwIUbZ6u0Wug54sLc=
+go.opentelemetry.io/otel/exporters/otlp/otlptrace/otlptracegrpc v1.43.0/go.mod h1:AGmbycVGEsRx9mXMZ75CsOyhSP6MFIcj/6dnG+vhVjk=
+go.opentelemetry.io/otel/exporters/prometheus v0.65.0 h1:jOveH/b4lU9HT7y+Gfamf18BqlOuz2PWEvs8yM7Q6XE=
+go.opentelemetry.io/otel/exporters/prometheus v0.65.0/go.mod h1:i1P8pcumauPtUI4YNopea1dhzEMuEqWP1xoUZDylLHo=
+go.opentelemetry.io/otel/metric v1.43.0 h1:d7638QeInOnuwOONPp4JAOGfbCEpYb+K6DVWvdxGzgM=
+go.opentelemetry.io/otel/metric v1.43.0/go.mod h1:RDnPtIxvqlgO8GRW18W6Z/4P462ldprJtfxHxyKd2PY=
+go.opentelemetry.io/otel/sdk v1.43.0 h1:pi5mE86i5rTeLXqoF/hhiBtUNcrAGHLKQdhg4h4V9Dg=
+go.opentelemetry.io/otel/sdk v1.43.0/go.mod h1:P+IkVU3iWukmiit/Yf9AWvpyRDlUeBaRg6Y+C58QHzg=
+go.opentelemetry.io/otel/sdk/metric v1.43.0 h1:S88dyqXjJkuBNLeMcVPRFXpRw2fuwdvfCGLEo89fDkw=
+go.opentelemetry.io/otel/sdk/metric v1.43.0/go.mod h1:C/RJtwSEJ5hzTiUz5pXF1kILHStzb9zFlIEe85bhj6A=
+go.opentelemetry.io/otel/trace v1.43.0 h1:BkNrHpup+4k4w+ZZ86CZoHHEkohws8AY+WTX09nk+3A=
+go.opentelemetry.io/otel/trace v1.43.0/go.mod h1:/QJhyVBUUswCphDVxq+8mld+AvhXZLhe+8WVFxiFff0=
+go.opentelemetry.io/proto/otlp v1.10.0 h1:IQRWgT5srOCYfiWnpqUYz9CVmbO8bFmKcwYxpuCSL2g=
+go.opentelemetry.io/proto/otlp v1.10.0/go.mod h1:/CV4QoCR/S9yaPj8utp3lvQPoqMtxXdzn7ozvvozVqk=
+go.uber.org/goleak v1.3.0 h1:2K3zAYmnTNqV73imy9J1T3WC+gmCePx2hEGkimedGto=
+go.uber.org/goleak v1.3.0/go.mod h1:CoHD4mav9JJNrW/WLlf7HGZPjdw8EucARQHekz1X6bE=
+go.yaml.in/yaml/v2 v2.4.4 h1:tuyd0P+2Ont/d6e2rl3be67goVK4R6deVxCUX5vyPaQ=
+go.yaml.in/yaml/v2 v2.4.4/go.mod h1:gMZqIpDtDqOfM0uNfy0SkpRhvUryYH0Z6wdMYcacYXQ=
+golang.org/x/crypto v0.52.0 h1:RMs7fP2rXdep0CftQlK8Uf+kibLm7qkCcradZWYz988=
+golang.org/x/crypto v0.52.0/go.mod h1:1QgfPxDqh0T2M/elOJtp9RvuR95kVjir0e6/BvEmGbc=
+golang.org/x/exp v0.0.0-20251113190631-e25ba8c21ef6 h1:zfMcR1Cs4KNuomFFgGefv5N0czO2XZpUbxGUy8i8ug0=
+golang.org/x/exp v0.0.0-20251113190631-e25ba8c21ef6/go.mod h1:46edojNIoXTNOhySWIWdix628clX9ODXwPsQuG6hsK0=
+golang.org/x/net v0.55.0 h1:bcvxaJn3e1U6InsFWt1JUq1aSjnRxLzT2rtD2KfkDF8=
+golang.org/x/net v0.55.0/go.mod h1:L5U2KuzuOe1lY7Z+aWVIKK6qEeJXnXV9yzGA+WCHJww=
+golang.org/x/sys v0.2.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
+golang.org/x/sys v0.10.0/go.mod h1:oPkhp1MJrh7nUepCBck5+mAzfO9JrbApNNgaTdGDITg=
+golang.org/x/sys v0.45.0 h1:dO4czNzziLiiXplLQgBCEpCvXQ3dnkn0SdaZSYdQ+FY=
+golang.org/x/sys v0.45.0/go.mod h1:4GL1E5IUh+htKOUEOaiffhrAeqysfVGipDYzABqnCmw=
+golang.org/x/term v0.43.0 h1:S4RLU2sB31O/NCl+zFN9Aru9A/Cq2aqKpTZJ6B+DwT4=
+golang.org/x/term v0.43.0/go.mod h1:lrhlHNdQJHO+1qVYiHfFKVuVioJIheAc3fBSMFYEIsk=
+golang.org/x/text v0.37.0 h1:Cqjiwd9eSg8e0QAkyCaQTNHFIIzWtidPahFWR83rTrc=
+golang.org/x/text v0.37.0/go.mod h1:a5sjxXGs9hsn/AJVwuElvCAo9v8QYLzvavO5z2PiM38=
+golang.org/x/time v0.12.0 h1:ScB/8o8olJvc+CQPWrK3fPZNfh7qgwCrY0zJmoEQLSE=
+golang.org/x/time v0.12.0/go.mod h1:CDIdPxbZBQxdj6cxyCIdrNogrJKMJ7pr37NYpMcMDSg=
 golang.zx2c4.com/wintun v0.0.0-20230126152724-0fa3db229ce2 h1:B82qJJgjvYKsXS9jeunTOisW56dUokqW/FOteYJJ/yg=
 golang.zx2c4.com/wintun v0.0.0-20230126152724-0fa3db229ce2/go.mod h1:deeaetjYA+DHMHg+sMSMI58GrEteJUUzzw7en6TJQcI=
-golang.zx2c4.com/wireguard v0.0.0-20231211153847-12269c276173 h1:/jFs0duh4rdb8uIfPMv78iAJGcPKDeqAFnaLBropIC4=
-golang.zx2c4.com/wireguard v0.0.0-20231211153847-12269c276173/go.mod h1:tkCQ4FQXmpAgYVh++1cq16/dH4QJtmvpRv19DWGAHSA=
-golang.zx2c4.com/wireguard/wgctrl v0.0.0-20230429144221-925a1e7659e6 h1:CawjfCvYQH2OU3/TnxLx97WDSUDRABfT18pCOYwc2GE=
-golang.zx2c4.com/wireguard/wgctrl v0.0.0-20230429144221-925a1e7659e6/go.mod h1:3rxYc4HtVcSG9gVaTs2GEBdehh+sYPOwKtyUWEOTb80=
-gvisor.dev/gvisor v0.0.0-20230927004350-cbd86285d259 h1:TbRPT0HtzFP3Cno1zZo7yPzEEnfu8EjLfl6IU9VfqkQ=
-gvisor.dev/gvisor v0.0.0-20230927004350-cbd86285d259/go.mod h1:AVgIgHMwK63XvmAzWG9vLQ41YnVHN0du0tEC46fI7yY=
+golang.zx2c4.com/wireguard v0.0.0-20250521234502-f333402bd9cb h1:whnFRlWMcXI9d+ZbWg+4sHnLp52d5yiIPUxMBSt4X9A=
+golang.zx2c4.com/wireguard v0.0.0-20250521234502-f333402bd9cb/go.mod h1:rpwXGsirqLqN2L0JDJQlwOboGHmptD5ZD6T2VmcqhTw=
+golang.zx2c4.com/wireguard/wgctrl v0.0.0-20241231184526-a9ab2273dd10 h1:3GDAcqdIg1ozBNLgPy4SLT84nfcBjr6rhGtXYtrkWLU=
+golang.zx2c4.com/wireguard/wgctrl v0.0.0-20241231184526-a9ab2273dd10/go.mod h1:T97yPqesLiNrOYxkwmhMI0ZIlJDm+p0PMR8eRVeR5tQ=
+golang.zx2c4.com/wireguard/windows v1.0.1 h1:eOxiDVbywPC+ZQqvdCK7x+ZwWXKbYv50TtH8ysFIbw8=
+golang.zx2c4.com/wireguard/windows v1.0.1/go.mod h1:+fbT3FFdX4zzYDLwJh5+HPEcNN/3HyNdzhNSVsQM+zs=
+gonum.org/v1/gonum v0.17.0 h1:VbpOemQlsSMrYmn7T2OUvQ4dqxQXU+ouZFQsZOx50z4=
+gonum.org/v1/gonum v0.17.0/go.mod h1:El3tOrEuMpv2UdMrbNlKEh9vd86bmQ6vqIcDwxEOc1E=
+google.golang.org/genproto/googleapis/api v0.0.0-20260401024825-9d38bb4040a9 h1:VPWxll4HlMw1Vs/qXtN7BvhZqsS9cdAittCNvVENElA=
+google.golang.org/genproto/googleapis/api v0.0.0-20260401024825-9d38bb4040a9/go.mod h1:7QBABkRtR8z+TEnmXTqIqwJLlzrZKVfAUm7tY3yGv0M=
+google.golang.org/genproto/googleapis/rpc v0.0.0-20260401024825-9d38bb4040a9 h1:m8qni9SQFH0tJc1X0vmnpw/0t+AImlSvp30sEupozUg=
+google.golang.org/genproto/googleapis/rpc v0.0.0-20260401024825-9d38bb4040a9/go.mod h1:4Hqkh8ycfw05ld/3BWL7rJOSfebL2Q+DVDeRgYgxUU8=
+google.golang.org/grpc v1.81.1 h1:VnnIIZ88UzOOKLukQi+ImGz8O1Wdp8nAGGnvOfEIWQQ=
+google.golang.org/grpc v1.81.1/go.mod h1:xGH9GfzOyMTGIOXBJmXt+BX/V0kcdQbdcuwQ/zNw42I=
+google.golang.org/protobuf v1.36.11 h1:fV6ZwhNocDyBLK0dj+fg8ektcVegBBuEolpbTQyBNVE=
+google.golang.org/protobuf v1.36.11/go.mod h1:HTf+CrKn2C3g5S8VImy6tdcUvCska2kB7j23XfzDpco=
+gopkg.in/check.v1 v0.0.0-20161208181325-20d25e280405/go.mod h1:Co6ibVJAznAaIkqp8huTwlJQCZ016jof/cbN4VW5Yz0=
+gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c h1:Hei/4ADfdWqJk1ZMxUNpqntNwaWcugrBjAiHlqqRiVk=
+gopkg.in/check.v1 v1.0.0-20201130134442-10cb98267c6c/go.mod h1:JHkPIbrfpd72SG/EVd6muEfDQjcINNoR0C8j2r3qZ4Q=
+gopkg.in/yaml.v3 v3.0.1 h1:fxVm/GzAzEWqLHuvctI91KS9hhNmmWOoWu0XTYJS7CA=
+gopkg.in/yaml.v3 v3.0.1/go.mod h1:K4uyk7z7BCEPqu6E+C64Yfv1cQ7kz7rIZviUmN+EgEM=
+gotest.tools/v3 v3.5.2 h1:7koQfIKdy+I8UTetycgUqXWSDwpgv193Ka+qRsmBY8Q=
+gotest.tools/v3 v3.5.2/go.mod h1:LtdLGcnqToBH83WByAAi/wiwSFCArdFIUV/xxN4pcjA=
+gvisor.dev/gvisor v0.0.0-20250503011706-39ed1f5ac29c h1:m/r7OM+Y2Ty1sgBQ7Qb27VgIMBW8ZZhT4gLnUyDIhzI=
+gvisor.dev/gvisor v0.0.0-20250503011706-39ed1f5ac29c/go.mod h1:3r5CMtNQMKIvBlrmM9xWUNamjKBYPOWyXOjmg5Kts3g=
+pgregory.net/rapid v1.2.0 h1:keKAYRcjm+e1F0oAuU5F5+YPAWcyxNNRK2wud503Gnk=
+pgregory.net/rapid v1.2.0/go.mod h1:PY5XlDGj0+V1FCq0o192FdRhpKHGTRIWBgqjDBTrq04=
+software.sslmate.com/src/go-pkcs12 v0.7.1 h1:bxkUPRsvTPNRBZa4M/aSX4PyMOEbq3V8I6hbkG4F4Q8=
+software.sslmate.com/src/go-pkcs12 v0.7.1/go.mod h1:Qiz0EyvDRJjjxGyUQa2cCNZn/wMyzrRJ/qcDXOQazLI=

healthcheck/healthcheck.go

@@ -0,0 +1,681 @@
+package healthcheck
+
+import (
+	"context"
+	"crypto/tls"
+	"encoding/json"
+	"fmt"
+	"net"
+	"net/http"
+	"strconv"
+	"strings"
+	"sync"
+	"time"
+
+	"github.com/fosrl/newt/logger"
+)
+
+// Health represents the health status of a target
+type Health int
+
+const (
+	StatusUnknown Health = iota
+	StatusHealthy
+	StatusUnhealthy
+)
+
+func (s Health) String() string {
+	switch s {
+	case StatusHealthy:
+		return "healthy"
+	case StatusUnhealthy:
+		return "unhealthy"
+	default:
+		return "unknown"
+	}
+}
+
+// Config holds the health check configuration for a target
+type Config struct {
+	ID                 int               `json:"id"`
+	Enabled            bool              `json:"hcEnabled"`
+	Path               string            `json:"hcPath"`
+	Scheme             string            `json:"hcScheme"`
+	Mode               string            `json:"hcMode"`
+	Hostname           string            `json:"hcHostname"`
+	Port               int               `json:"hcPort"`
+	Interval           int               `json:"hcInterval"`          // in seconds
+	UnhealthyInterval  int               `json:"hcUnhealthyInterval"` // in seconds
+	Timeout            int               `json:"hcTimeout"`           // in seconds
+	FollowRedirects    *bool             `json:"hcFollowRedirects"`
+	Headers            map[string]string `json:"hcHeaders"`
+	Method             string            `json:"hcMethod"`
+	Status             int               `json:"hcStatus"` // HTTP status code
+	TLSServerName      string            `json:"hcTlsServerName"`
+	HealthyThreshold   int               `json:"hcHealthyThreshold"`   // consecutive successes required to become healthy
+	UnhealthyThreshold int               `json:"hcUnhealthyThreshold"` // consecutive failures required to become unhealthy
+}
+
+// Target represents a health check target with its current status
+type Target struct {
+	Config               Config    `json:"config"`
+	Status               Health    `json:"status"`
+	LastCheck            time.Time `json:"lastCheck"`
+	LastError            string    `json:"lastError,omitempty"`
+	CheckCount           int       `json:"checkCount"`
+	timer                *time.Timer
+	ctx                  context.Context
+	cancel               context.CancelFunc
+	client               *http.Client
+	consecutiveSuccesses int
+	consecutiveFailures  int
+}
+
+// StatusChangeCallback is called when any target's status changes
+type StatusChangeCallback func(targets map[int]*Target)
+
+// Monitor manages health check targets and their monitoring
+type Monitor struct {
+	targets     map[int]*Target
+	mutex       sync.RWMutex
+	callback    StatusChangeCallback
+	enforceCert bool
+}
+
+// NewMonitor creates a new health check monitor
+func NewMonitor(callback StatusChangeCallback, enforceCert bool) *Monitor {
+	logger.Debug("Creating new health check monitor with certificate enforcement: %t", enforceCert)
+
+	return &Monitor{
+		targets:     make(map[int]*Target),
+		callback:    callback,
+		enforceCert: enforceCert,
+	}
+}
+
+// parseHeaders parses the headers string into a map
+func parseHeaders(headersStr string) map[string]string {
+	headers := make(map[string]string)
+	if headersStr == "" {
+		return headers
+	}
+
+	// Try to parse as JSON first
+	if err := json.Unmarshal([]byte(headersStr), &headers); err == nil {
+		return headers
+	}
+
+	// Fallback to simple key:value parsing
+	pairs := strings.Split(headersStr, ",")
+	for _, pair := range pairs {
+		kv := strings.SplitN(strings.TrimSpace(pair), ":", 2)
+		if len(kv) == 2 {
+			headers[strings.TrimSpace(kv[0])] = strings.TrimSpace(kv[1])
+		}
+	}
+	return headers
+}
+
+// AddTarget adds a new health check target
+func (m *Monitor) AddTarget(config Config) error {
+	m.mutex.Lock()
+	defer m.mutex.Unlock()
+
+	logger.Info("Adding health check target: ID=%d, hostname=%s, port=%d, enabled=%t",
+		config.ID, config.Hostname, config.Port, config.Enabled)
+
+	return m.addTargetUnsafe(config)
+}
+
+// AddTargets adds multiple health check targets in bulk
+func (m *Monitor) AddTargets(configs []Config) error {
+	m.mutex.Lock()
+	defer m.mutex.Unlock()
+
+	logger.Debug("Adding %d health check targets in bulk", len(configs))
+
+	for _, config := range configs {
+		if err := m.addTargetUnsafe(config); err != nil {
+			logger.Error("Failed to add target %d: %v", config.ID, err)
+			return fmt.Errorf("failed to add target %d: %v", config.ID, err)
+		}
+		logger.Debug("Successfully added target: ID=%d, hostname=%s", config.ID, config.Hostname)
+	}
+
+	// Don't notify callback immediately - let the initial health checks complete first
+	// The callback will be triggered when the first health check results are available
+
+	logger.Debug("Successfully added all %d health check targets", len(configs))
+	return nil
+}
+
+// addTargetUnsafe adds a target without acquiring the mutex (internal method)
+func (m *Monitor) addTargetUnsafe(config Config) error {
+	// Set defaults
+	if config.Scheme == "" {
+		config.Scheme = "http"
+	}
+	if config.Mode == "" {
+		config.Mode = "http"
+	}
+	if config.Method == "" {
+		config.Method = "GET"
+	}
+	if config.Interval == 0 {
+		config.Interval = 30
+	}
+	if config.UnhealthyInterval == 0 {
+		config.UnhealthyInterval = 30
+	}
+	if config.Timeout == 0 {
+		config.Timeout = 5
+	}
+	if config.HealthyThreshold == 0 {
+		config.HealthyThreshold = 1
+	}
+	if config.UnhealthyThreshold == 0 {
+		config.UnhealthyThreshold = 1
+	}
+
+	logger.Debug("Target %d configuration: mode=%s, scheme=%s, method=%s, interval=%ds, timeout=%ds, healthyThreshold=%d, unhealthyThreshold=%d",
+		config.ID, config.Mode, config.Scheme, config.Method, config.Interval, config.Timeout,
+		config.HealthyThreshold, config.UnhealthyThreshold)
+
+	// Parse headers if provided as string
+	if len(config.Headers) == 0 && config.Path != "" {
+		// This is a simplified header parsing - in real use you might want more robust parsing
+		config.Headers = make(map[string]string)
+	}
+
+	// Remove existing target if it exists
+	if existing, exists := m.targets[config.ID]; exists {
+		logger.Info("Replacing existing target with ID %d", config.ID)
+		existing.cancel()
+	}
+
+	// Create new target
+	ctx, cancel := context.WithCancel(context.Background())
+	target := &Target{
+		Config: config,
+		Status: StatusUnknown,
+		ctx:    ctx,
+		cancel: cancel,
+		client: &http.Client{
+			CheckRedirect: func() func(*http.Request, []*http.Request) error {
+				// Default to following redirects if not explicitly configured
+				followRedirects := config.FollowRedirects == nil || *config.FollowRedirects
+				if !followRedirects {
+					return func(req *http.Request, via []*http.Request) error {
+						return http.ErrUseLastResponse
+					}
+				}
+				return nil
+			}(),
+			Transport: &http.Transport{
+				// Disable keep-alives so each health check uses a fresh connection.
+				// Reusing idle connections causes spurious timeouts when the remote
+				// server closes the connection between long check intervals.
+				DisableKeepAlives: true,
+				TLSClientConfig: &tls.Config{
+					// Configure TLS settings based on certificate enforcement
+					InsecureSkipVerify: !m.enforceCert,
+					// Use SNI TLS header if present
+					ServerName: config.TLSServerName,
+				},
+			},
+		},
+	}
+
+	m.targets[config.ID] = target
+
+	// Start monitoring if enabled
+	if config.Enabled {
+		logger.Info("Starting monitoring for target %d (%s:%d)", config.ID, config.Hostname, config.Port)
+		go m.monitorTarget(target)
+	} else {
+		logger.Debug("Target %d added but monitoring is disabled", config.ID)
+	}
+
+	return nil
+}
+
+// RemoveTarget removes a health check target
+func (m *Monitor) RemoveTarget(id int) error {
+	m.mutex.Lock()
+	defer m.mutex.Unlock()
+
+	target, exists := m.targets[id]
+	if !exists {
+		logger.Warn("Attempted to remove non-existent target with ID %d", id)
+		return fmt.Errorf("target with id %d not found", id)
+	}
+
+	logger.Info("Removing health check target: ID=%d", id)
+	target.cancel()
+	delete(m.targets, id)
+
+	// Notify callback of status change
+	if m.callback != nil {
+		go m.callback(m.getAllTargetsUnsafe())
+	}
+
+	logger.Info("Successfully removed target %d", id)
+	return nil
+}
+
+// RemoveTargets removes multiple health check targets
+func (m *Monitor) RemoveTargets(ids []int) error {
+	m.mutex.Lock()
+	defer m.mutex.Unlock()
+
+	logger.Info("Removing %d health check targets", len(ids))
+	var notFound []int
+
+	for _, id := range ids {
+		target, exists := m.targets[id]
+		if !exists {
+			notFound = append(notFound, id)
+			logger.Warn("Target with ID %d not found during bulk removal", id)
+			continue
+		}
+
+		logger.Debug("Removing target %d", id)
+		target.cancel()
+		delete(m.targets, id)
+	}
+
+	removedCount := len(ids) - len(notFound)
+	logger.Info("Successfully removed %d targets", removedCount)
+
+	// Notify callback of status change if any targets were removed
+	if len(notFound) != len(ids) && m.callback != nil {
+		go m.callback(m.getAllTargetsUnsafe())
+	}
+
+	if len(notFound) > 0 {
+		logger.Error("Some targets not found during removal: %v", notFound)
+		return fmt.Errorf("targets not found: %v", notFound)
+	}
+
+	return nil
+}
+
+// RemoveTargetsByID is a convenience method that accepts either a single ID or multiple IDs
+func (m *Monitor) RemoveTargetsByID(ids ...int) error {
+	return m.RemoveTargets(ids)
+}
+
+// GetTargets returns a copy of all targets
+func (m *Monitor) GetTargets() map[int]*Target {
+	m.mutex.RLock()
+	defer m.mutex.RUnlock()
+	return m.getAllTargetsUnsafe()
+}
+
+// getAllTargetsUnsafe returns a copy of all targets without acquiring the mutex (internal method)
+func (m *Monitor) getAllTargetsUnsafe() map[int]*Target {
+	targets := make(map[int]*Target)
+	for id, target := range m.targets {
+		// Create a copy to avoid race conditions
+		targetCopy := *target
+		targets[id] = &targetCopy
+	}
+	return targets
+}
+
+// getAllTargets returns a copy of all targets (deprecated, use GetTargets)
+func (m *Monitor) getAllTargets() map[int]*Target {
+	return m.GetTargets()
+}
+
+// monitorTarget monitors a single target
+func (m *Monitor) monitorTarget(target *Target) {
+	logger.Info("Starting health check monitoring for target %d (%s:%d)",
+		target.Config.ID, target.Config.Hostname, target.Config.Port)
+
+	// Initial check
+	oldStatus := target.Status
+	m.performHealthCheck(target)
+
+	// Notify callback after initial check if status changed or if it's the first check
+	if (oldStatus != target.Status || oldStatus == StatusUnknown) && m.callback != nil {
+		logger.Info("Target %d initial status: %s", target.Config.ID, target.Status.String())
+		go m.callback(m.GetTargets())
+	}
+
+	// Set up timer based on current status
+	interval := time.Duration(target.Config.Interval) * time.Second
+	if target.Status == StatusUnhealthy {
+		interval = time.Duration(target.Config.UnhealthyInterval) * time.Second
+	}
+
+	logger.Debug("Target %d: initial check interval set to %v", target.Config.ID, interval)
+	target.timer = time.NewTimer(interval)
+	defer target.timer.Stop()
+
+	for {
+		select {
+		case <-target.ctx.Done():
+			logger.Info("Stopping health check monitoring for target %d", target.Config.ID)
+			return
+		case <-target.timer.C:
+			oldStatus := target.Status
+			m.performHealthCheck(target)
+
+			// Update timer interval if status changed
+			newInterval := time.Duration(target.Config.Interval) * time.Second
+			if target.Status == StatusUnhealthy {
+				newInterval = time.Duration(target.Config.UnhealthyInterval) * time.Second
+			}
+
+			if newInterval != interval {
+				logger.Debug("Target %d: updating check interval from %v to %v due to status change",
+					target.Config.ID, interval, newInterval)
+				interval = newInterval
+			}
+
+			// Reset timer for next check with current interval
+			target.timer.Reset(interval)
+
+			// Notify callback if status changed
+			if oldStatus != target.Status && m.callback != nil {
+				logger.Info("Target %d status changed: %s -> %s",
+					target.Config.ID, oldStatus.String(), target.Status.String())
+				go m.callback(m.GetTargets())
+			}
+		}
+	}
+}
+
+// performHealthCheck performs a health check on a target and applies threshold logic
+func (m *Monitor) performHealthCheck(target *Target) {
+	target.CheckCount++
+	target.LastCheck = time.Now()
+
+	var passed bool
+	var checkErr string
+
+	switch strings.ToLower(target.Config.Mode) {
+	case "tcp":
+		passed, checkErr = m.performTCPCheck(target)
+	default:
+		// "http", "https", or anything else falls through to HTTP
+		passed, checkErr = m.performHTTPCheck(target)
+	}
+
+	if passed {
+		target.consecutiveFailures = 0
+		target.consecutiveSuccesses++
+
+		logger.Debug("Target %d: check passed (consecutive successes: %d / threshold: %d)",
+			target.Config.ID, target.consecutiveSuccesses, target.Config.HealthyThreshold)
+
+		if target.consecutiveSuccesses >= target.Config.HealthyThreshold {
+			target.Status = StatusHealthy
+			target.LastError = ""
+		}
+	} else {
+		target.consecutiveSuccesses = 0
+		target.consecutiveFailures++
+		target.LastError = checkErr
+
+		logger.Debug("Target %d: check failed (consecutive failures: %d / threshold: %d): %s",
+			target.Config.ID, target.consecutiveFailures, target.Config.UnhealthyThreshold, checkErr)
+
+		if target.consecutiveFailures >= target.Config.UnhealthyThreshold {
+			target.Status = StatusUnhealthy
+		}
+	}
+}
+
+// performTCPCheck dials the target's host:port over TCP and returns whether it succeeded
+func (m *Monitor) performTCPCheck(target *Target) (bool, string) {
+	address := net.JoinHostPort(target.Config.Hostname, strconv.Itoa(target.Config.Port))
+	timeout := time.Duration(target.Config.Timeout) * time.Second
+
+	logger.Debug("Target %d: performing TCP health check to %s (timeout: %v)",
+		target.Config.ID, address, timeout)
+
+	ctx, cancel := context.WithTimeout(target.ctx, timeout)
+	defer cancel()
+	conn, err := (&net.Dialer{}).DialContext(ctx, "tcp", address)
+	if err != nil {
+		msg := fmt.Sprintf("TCP dial failed: %v", err)
+		logger.Warn("Target %d: %s", target.Config.ID, msg)
+		return false, msg
+	}
+	conn.Close()
+
+	logger.Debug("Target %d: TCP health check passed", target.Config.ID)
+	return true, ""
+}
+
+// performHTTPCheck performs an HTTP/HTTPS health check and returns whether it succeeded
+func (m *Monitor) performHTTPCheck(target *Target) (bool, string) {
+	// Build URL (use net.JoinHostPort to properly handle IPv6 addresses with ports)
+	host := target.Config.Hostname
+	if target.Config.Port > 0 {
+		host = net.JoinHostPort(target.Config.Hostname, strconv.Itoa(target.Config.Port))
+	}
+	url := fmt.Sprintf("%s://%s", target.Config.Scheme, host)
+	if target.Config.Path != "" {
+		if !strings.HasPrefix(target.Config.Path, "/") {
+			url += "/"
+		}
+		url += target.Config.Path
+	}
+
+	logger.Debug("Target %d: performing HTTP health check %d to %s",
+		target.Config.ID, target.CheckCount, url)
+
+	if target.Config.Scheme == "https" {
+		logger.Debug("Target %d: HTTPS health check with certificate enforcement: %t",
+			target.Config.ID, m.enforceCert)
+	}
+
+	// Create request with timeout context, derived from the target's context so
+	// that in-flight requests are cancelled immediately when the target is replaced.
+	ctx, cancel := context.WithTimeout(target.ctx, time.Duration(target.Config.Timeout)*time.Second)
+	defer cancel()
+
+	req, err := http.NewRequestWithContext(ctx, target.Config.Method, url, nil)
+	if err != nil {
+		msg := fmt.Sprintf("failed to create request: %v", err)
+		logger.Warn("Target %d: %s", target.Config.ID, msg)
+		return false, msg
+	}
+
+	// Add headers
+	for key, value := range target.Config.Headers {
+		// Handle Host header specially - it must be set on req.Host, not in headers
+		if strings.EqualFold(key, "Host") {
+			req.Host = value
+		} else {
+			req.Header.Set(key, value)
+		}
+	}
+
+	// Perform request
+	resp, err := target.client.Do(req)
+	if err != nil {
+		msg := fmt.Sprintf("request failed: %v", err)
+		logger.Warn("Target %d: health check failed: %v", target.Config.ID, err)
+		return false, msg
+	}
+	defer resp.Body.Close()
+
+	// Check response status
+	if target.Config.Status > 0 {
+		// Check for specific status code
+		logger.Debug("Target %d: checking status against expected code %d", target.Config.ID, target.Config.Status)
+		if resp.StatusCode == target.Config.Status {
+			logger.Debug("Target %d: health check passed (status: %d)", target.Config.ID, resp.StatusCode)
+			return true, ""
+		}
+		msg := fmt.Sprintf("unexpected status code: %d (expected: %d)", resp.StatusCode, target.Config.Status)
+		logger.Warn("Target %d: %s", target.Config.ID, msg)
+		return false, msg
+	}
+
+	// Default: check for 2xx range
+	if resp.StatusCode >= 200 && resp.StatusCode < 300 {
+		logger.Debug("Target %d: health check passed (status: %d)", target.Config.ID, resp.StatusCode)
+		return true, ""
+	}
+
+	msg := fmt.Sprintf("unhealthy status code: %d", resp.StatusCode)
+	logger.Warn("Target %d: health check failed with status code %d", target.Config.ID, resp.StatusCode)
+	return false, msg
+}
+
+// Stop stops monitoring all targets
+func (m *Monitor) Stop() {
+	m.mutex.Lock()
+	defer m.mutex.Unlock()
+
+	targetCount := len(m.targets)
+	logger.Info("Stopping health check monitor with %d targets", targetCount)
+
+	for id, target := range m.targets {
+		logger.Debug("Stopping monitoring for target %d", id)
+		target.cancel()
+	}
+	m.targets = make(map[int]*Target)
+
+	logger.Info("Health check monitor stopped")
+}
+
+// EnableTarget enables monitoring for a specific target
+func (m *Monitor) EnableTarget(id int) error {
+	m.mutex.Lock()
+	defer m.mutex.Unlock()
+
+	target, exists := m.targets[id]
+	if !exists {
+		logger.Warn("Attempted to enable non-existent target with ID %d", id)
+		return fmt.Errorf("target with id %d not found", id)
+	}
+
+	if !target.Config.Enabled {
+		logger.Info("Enabling health check monitoring for target %d", id)
+		target.Config.Enabled = true
+		target.cancel() // Stop existing monitoring
+
+		ctx, cancel := context.WithCancel(context.Background())
+		target.ctx = ctx
+		target.cancel = cancel
+
+		go m.monitorTarget(target)
+	} else {
+		logger.Debug("Target %d is already enabled", id)
+	}
+
+	return nil
+}
+
+// DisableTarget disables monitoring for a specific target
+func (m *Monitor) DisableTarget(id int) error {
+	m.mutex.Lock()
+	defer m.mutex.Unlock()
+
+	target, exists := m.targets[id]
+	if !exists {
+		logger.Warn("Attempted to disable non-existent target with ID %d", id)
+		return fmt.Errorf("target with id %d not found", id)
+	}
+
+	if target.Config.Enabled {
+		logger.Info("Disabling health check monitoring for target %d", id)
+		target.Config.Enabled = false
+		target.cancel()
+		target.Status = StatusUnknown
+
+		// Notify callback of status change
+		if m.callback != nil {
+			go m.callback(m.getAllTargetsUnsafe())
+		}
+	} else {
+		logger.Debug("Target %d is already disabled", id)
+	}
+
+	return nil
+}
+
+// GetTargetIDs returns a slice of all current target IDs
+func (m *Monitor) GetTargetIDs() []int {
+	m.mutex.RLock()
+	defer m.mutex.RUnlock()
+
+	ids := make([]int, 0, len(m.targets))
+	for id := range m.targets {
+		ids = append(ids, id)
+	}
+	return ids
+}
+
+// SyncTargets synchronizes the current targets to match the desired set.
+// It removes targets not in the desired set and adds targets that are missing.
+func (m *Monitor) SyncTargets(desiredConfigs []Config) error {
+	m.mutex.Lock()
+	defer m.mutex.Unlock()
+
+	logger.Info("Syncing health check targets: %d desired targets", len(desiredConfigs))
+
+	// Build a set of desired target IDs
+	desiredIDs := make(map[int]Config)
+	for _, config := range desiredConfigs {
+		desiredIDs[config.ID] = config
+	}
+
+	// Find targets to remove (exist but not in desired set)
+	var toRemove []int
+	for id := range m.targets {
+		if _, exists := desiredIDs[id]; !exists {
+			toRemove = append(toRemove, id)
+		}
+	}
+
+	// Remove targets that are not in the desired set
+	for _, id := range toRemove {
+		logger.Info("Sync: removing health check target %d", id)
+		if target, exists := m.targets[id]; exists {
+			target.cancel()
+			delete(m.targets, id)
+		}
+	}
+
+	// Add or update targets from the desired set
+	var addedCount, updatedCount int
+	for id, config := range desiredIDs {
+		if existing, exists := m.targets[id]; exists {
+			// Target exists - check if config changed and update if needed
+			// For now, we'll replace it to ensure config is up to date
+			logger.Debug("Sync: updating health check target %d", id)
+			existing.cancel()
+			delete(m.targets, id)
+			if err := m.addTargetUnsafe(config); err != nil {
+				logger.Error("Sync: failed to update target %d: %v", id, err)
+				return fmt.Errorf("failed to update target %d: %v", id, err)
+			}
+			updatedCount++
+		} else {
+			// Target doesn't exist - add it
+			logger.Debug("Sync: adding health check target %d", id)
+			if err := m.addTargetUnsafe(config); err != nil {
+				logger.Error("Sync: failed to add target %d: %v", id, err)
+				return fmt.Errorf("failed to add target %d: %v", id, err)
+			}
+			addedCount++
+		}
+	}
+
+	logger.Info("Sync complete: removed %d, added %d, updated %d targets",
+		len(toRemove), addedCount, updatedCount)
+
+	// Notify callback if any changes were made
+	if (len(toRemove) > 0 || addedCount > 0 || updatedCount > 0) && m.callback != nil {
+		go m.callback(m.getAllTargetsUnsafe())
+	}
+
+	return nil
+}

holepunch/holepunch.go

@@ -0,0 +1,624 @@
+package holepunch
+
+import (
+	"encoding/json"
+	"fmt"
+	"net"
+	"strconv"
+	"sync"
+	"time"
+
+	"github.com/fosrl/newt/bind"
+	"github.com/fosrl/newt/logger"
+	"github.com/fosrl/newt/util"
+	"golang.org/x/crypto/chacha20poly1305"
+	"golang.org/x/crypto/curve25519"
+	mrand "golang.org/x/exp/rand"
+	"golang.zx2c4.com/wireguard/wgctrl/wgtypes"
+)
+
+// ExitNode represents a WireGuard exit node for hole punching
+type ExitNode struct {
+	Endpoint  string `json:"endpoint"`
+	RelayPort uint16 `json:"relayPort"`
+	PublicKey string `json:"publicKey"`
+	SiteIds   []int  `json:"siteIds,omitempty"`
+}
+
+// Manager handles UDP hole punching operations
+type Manager struct {
+	mu         sync.Mutex
+	running    bool
+	stopChan   chan struct{}
+	sharedBind *bind.SharedBind
+	ID         string
+	token      string
+	publicKey  string
+	clientType string
+	exitNodes  map[string]ExitNode // key is endpoint
+	updateChan chan struct{}       // signals the goroutine to refresh exit nodes
+	publicDNS  []string
+
+	sendHolepunchInterval    time.Duration
+	sendHolepunchIntervalMin time.Duration
+	sendHolepunchIntervalMax time.Duration
+	defaultIntervalMin       time.Duration
+	defaultIntervalMax       time.Duration
+}
+
+const defaultSendHolepunchIntervalMax = 60 * time.Second
+const defaultSendHolepunchIntervalMin = 1 * time.Second
+
+// NewManager creates a new hole punch manager
+func NewManager(sharedBind *bind.SharedBind, ID string, clientType string, publicKey string, publicDNS []string) *Manager {
+	return &Manager{
+		sharedBind:               sharedBind,
+		ID:                       ID,
+		clientType:               clientType,
+		publicKey:                publicKey,
+		publicDNS:                publicDNS,
+		exitNodes:                make(map[string]ExitNode),
+		sendHolepunchInterval:    defaultSendHolepunchIntervalMin,
+		sendHolepunchIntervalMin: defaultSendHolepunchIntervalMin,
+		sendHolepunchIntervalMax: defaultSendHolepunchIntervalMax,
+		defaultIntervalMin:       defaultSendHolepunchIntervalMin,
+		defaultIntervalMax:       defaultSendHolepunchIntervalMax,
+	}
+}
+
+// SetToken updates the authentication token used for hole punching
+func (m *Manager) SetToken(token string) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	m.token = token
+}
+
+// SetPublicDNS replaces the list of DNS servers used to resolve exit-node
+// endpoints.  The servers must be in "host:port" format (e.g. "8.8.8.8:53").
+func (m *Manager) SetPublicDNS(servers []string) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	m.publicDNS = servers
+}
+
+// IsRunning returns whether hole punching is currently active
+func (m *Manager) IsRunning() bool {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	return m.running
+}
+
+// Stop stops any ongoing hole punch operations
+func (m *Manager) Stop() {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+
+	if !m.running {
+		return
+	}
+
+	if m.stopChan != nil {
+		close(m.stopChan)
+		m.stopChan = nil
+	}
+
+	if m.updateChan != nil {
+		close(m.updateChan)
+		m.updateChan = nil
+	}
+
+	m.running = false
+	logger.Info("Hole punch manager stopped")
+}
+
+// AddExitNode adds a new exit node to the rotation if it doesn't already exist
+func (m *Manager) AddExitNode(exitNode ExitNode) bool {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+
+	if _, exists := m.exitNodes[exitNode.Endpoint]; exists {
+		logger.Debug("Exit node %s already exists in rotation", exitNode.Endpoint)
+		return false
+	}
+
+	m.exitNodes[exitNode.Endpoint] = exitNode
+	logger.Info("Added exit node %s to hole punch rotation", exitNode.Endpoint)
+
+	// Signal the goroutine to refresh if running
+	if m.running && m.updateChan != nil {
+		select {
+		case m.updateChan <- struct{}{}:
+		default:
+			// Channel full or closed, skip
+		}
+	}
+
+	return true
+}
+
+// RemoveExitNode removes an exit node from the rotation
+func (m *Manager) RemoveExitNode(endpoint string) bool {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+
+	if _, exists := m.exitNodes[endpoint]; !exists {
+		logger.Debug("Exit node %s not found in rotation", endpoint)
+		return false
+	}
+
+	delete(m.exitNodes, endpoint)
+	logger.Info("Removed exit node %s from hole punch rotation", endpoint)
+
+	// Signal the goroutine to refresh if running
+	if m.running && m.updateChan != nil {
+		select {
+		case m.updateChan <- struct{}{}:
+		default:
+			// Channel full or closed, skip
+		}
+	}
+
+	return true
+}
+
+/*
+RemoveExitNodesByPeer removes the peer ID from the SiteIds list in each exit node.
+If the SiteIds list becomes empty after removal, the exit node is removed entirely.
+Returns the number of exit nodes removed.
+*/
+func (m *Manager) RemoveExitNodesByPeer(peerID int) int {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+
+	removed := 0
+	for endpoint, node := range m.exitNodes {
+		// Remove peerID from SiteIds if present
+		newSiteIds := make([]int, 0, len(node.SiteIds))
+		for _, id := range node.SiteIds {
+			if id != peerID {
+				newSiteIds = append(newSiteIds, id)
+			}
+		}
+		if len(newSiteIds) != len(node.SiteIds) {
+			node.SiteIds = newSiteIds
+			if len(node.SiteIds) == 0 {
+				delete(m.exitNodes, endpoint)
+				logger.Info("Removed exit node %s as no more site IDs remain after removing peer %d", endpoint, peerID)
+				removed++
+			} else {
+				m.exitNodes[endpoint] = node
+				logger.Info("Removed peer %d from exit node %s site IDs", peerID, endpoint)
+			}
+		}
+	}
+
+	if removed > 0 {
+		// Signal the goroutine to refresh if running
+		if m.running && m.updateChan != nil {
+			select {
+			case m.updateChan <- struct{}{}:
+			default:
+				// Channel full or closed, skip
+			}
+		}
+	}
+
+	return removed
+}
+
+// GetExitNodes returns a copy of the current exit nodes
+func (m *Manager) GetExitNodes() []ExitNode {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+
+	nodes := make([]ExitNode, 0, len(m.exitNodes))
+	for _, node := range m.exitNodes {
+		nodes = append(nodes, node)
+	}
+	return nodes
+}
+
+// SetServerHolepunchInterval sets custom min and max intervals for hole punching.
+// This is useful for low power mode where longer intervals are desired.
+func (m *Manager) SetServerHolepunchInterval(min, max time.Duration) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+
+	m.sendHolepunchIntervalMin = min
+	m.sendHolepunchIntervalMax = max
+	m.sendHolepunchInterval = min
+
+	logger.Info("Set hole punch intervals: min=%v, max=%v", min, max)
+
+	// Signal the goroutine to apply the new interval if running
+	if m.running && m.updateChan != nil {
+		select {
+		case m.updateChan <- struct{}{}:
+		default:
+			// Channel full or closed, skip
+		}
+	}
+}
+
+// GetInterval returns the current min and max intervals
+func (m *Manager) GetServerHolepunchInterval() (min, max time.Duration) {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+	return m.sendHolepunchIntervalMin, m.sendHolepunchIntervalMax
+}
+
+// ResetServerHolepunchInterval resets the hole punch interval back to the default values.
+// This restores normal operation after low power mode or other custom settings.
+func (m *Manager) ResetServerHolepunchInterval() {
+	m.mu.Lock()
+	defer m.mu.Unlock()
+
+	m.sendHolepunchIntervalMin = m.defaultIntervalMin
+	m.sendHolepunchIntervalMax = m.defaultIntervalMax
+	m.sendHolepunchInterval = m.defaultIntervalMin
+
+	logger.Info("Reset hole punch intervals to defaults: min=%v, max=%v", m.defaultIntervalMin, m.defaultIntervalMax)
+
+	// Signal the goroutine to apply the new interval if running
+	if m.running && m.updateChan != nil {
+		select {
+		case m.updateChan <- struct{}{}:
+		default:
+			// Channel full or closed, skip
+		}
+	}
+}
+
+// TriggerHolePunch sends an immediate hole punch packet to all configured exit nodes
+// This is useful for triggering hole punching on demand without waiting for the interval
+func (m *Manager) TriggerHolePunch() error {
+	m.mu.Lock()
+
+	if len(m.exitNodes) == 0 {
+		m.mu.Unlock()
+		return fmt.Errorf("no exit nodes configured")
+	}
+
+	// Get a copy of exit nodes to work with
+	currentExitNodes := make([]ExitNode, 0, len(m.exitNodes))
+	for _, node := range m.exitNodes {
+		currentExitNodes = append(currentExitNodes, node)
+	}
+	m.mu.Unlock()
+
+	logger.Info("Triggering on-demand hole punch to %d exit nodes", len(currentExitNodes))
+
+	// Send hole punch to all exit nodes
+	successCount := 0
+	for _, exitNode := range currentExitNodes {
+		var host string
+		var err error
+		if len(m.publicDNS) > 0 {
+			host, err = util.ResolveDomainUpstream(exitNode.Endpoint, m.publicDNS)
+		} else {
+			host, err = util.ResolveDomain(exitNode.Endpoint)
+		}
+		if err != nil {
+			logger.Warn("Failed to resolve endpoint %s: %v", exitNode.Endpoint, err)
+			continue
+		}
+
+		serverAddr := net.JoinHostPort(host, strconv.Itoa(int(exitNode.RelayPort)))
+		remoteAddr, err := net.ResolveUDPAddr("udp", serverAddr)
+		if err != nil {
+			logger.Error("Failed to resolve UDP address %s: %v", serverAddr, err)
+			continue
+		}
+
+		if err := m.sendHolePunch(remoteAddr, exitNode.PublicKey); err != nil {
+			logger.Warn("Failed to send on-demand hole punch to %s: %v", exitNode.Endpoint, err)
+			continue
+		}
+
+		logger.Debug("Sent on-demand hole punch to %s", exitNode.Endpoint)
+		successCount++
+	}
+
+	if successCount == 0 {
+		return fmt.Errorf("failed to send hole punch to any exit node")
+	}
+
+	logger.Info("Successfully sent on-demand hole punch to %d/%d exit nodes", successCount, len(currentExitNodes))
+	return nil
+}
+
+// StartMultipleExitNodes starts hole punching to multiple exit nodes
+func (m *Manager) StartMultipleExitNodes(exitNodes []ExitNode) error {
+	m.mu.Lock()
+
+	if m.running {
+		m.mu.Unlock()
+		logger.Debug("UDP hole punch already running, skipping new request")
+		return fmt.Errorf("hole punch already running")
+	}
+
+	// Populate exit nodes map
+	m.exitNodes = make(map[string]ExitNode)
+	for _, node := range exitNodes {
+		m.exitNodes[node.Endpoint] = node
+	}
+
+	m.running = true
+	m.stopChan = make(chan struct{})
+	m.updateChan = make(chan struct{}, 1)
+	m.mu.Unlock()
+
+	logger.Debug("Starting UDP hole punch to %d exit nodes with shared bind", len(exitNodes))
+
+	go m.runMultipleExitNodes()
+
+	return nil
+}
+
+// Start starts hole punching with the current set of exit nodes
+func (m *Manager) Start() error {
+	m.mu.Lock()
+
+	if m.running {
+		m.mu.Unlock()
+		logger.Debug("UDP hole punch already running")
+		return fmt.Errorf("hole punch already running")
+	}
+
+	m.running = true
+	m.stopChan = make(chan struct{})
+	m.updateChan = make(chan struct{}, 1)
+	nodeCount := len(m.exitNodes)
+	m.mu.Unlock()
+
+	if nodeCount == 0 {
+		logger.Info("Starting UDP hole punch manager (waiting for exit nodes to be added)")
+	} else {
+		logger.Info("Starting UDP hole punch with %d exit nodes", nodeCount)
+	}
+
+	go m.runMultipleExitNodes()
+
+	return nil
+}
+
+// runMultipleExitNodes performs hole punching to multiple exit nodes
+func (m *Manager) runMultipleExitNodes() {
+	defer func() {
+		m.mu.Lock()
+		m.running = false
+		m.mu.Unlock()
+		logger.Info("UDP hole punch goroutine ended for all exit nodes")
+	}()
+
+	// Resolve all endpoints upfront
+	type resolvedExitNode struct {
+		remoteAddr   *net.UDPAddr
+		publicKey    string
+		endpointName string
+	}
+
+	resolveNodes := func() []resolvedExitNode {
+		m.mu.Lock()
+		currentExitNodes := make([]ExitNode, 0, len(m.exitNodes))
+		for _, node := range m.exitNodes {
+			currentExitNodes = append(currentExitNodes, node)
+		}
+		m.mu.Unlock()
+
+		var resolvedNodes []resolvedExitNode
+		for _, exitNode := range currentExitNodes {
+			var host string
+			var err error
+			if len(m.publicDNS) > 0 {
+				host, err = util.ResolveDomainUpstream(exitNode.Endpoint, m.publicDNS)
+			} else {
+				host, err = util.ResolveDomain(exitNode.Endpoint)
+			}
+			if err != nil {
+				logger.Warn("Failed to resolve endpoint %s: %v", exitNode.Endpoint, err)
+				continue
+			}
+
+			serverAddr := net.JoinHostPort(host, strconv.Itoa(int(exitNode.RelayPort)))
+			remoteAddr, err := net.ResolveUDPAddr("udp", serverAddr)
+			if err != nil {
+				logger.Error("Failed to resolve UDP address %s: %v", serverAddr, err)
+				continue
+			}
+
+			resolvedNodes = append(resolvedNodes, resolvedExitNode{
+				remoteAddr:   remoteAddr,
+				publicKey:    exitNode.PublicKey,
+				endpointName: exitNode.Endpoint,
+			})
+			logger.Debug("Resolved exit node: %s -> %s", exitNode.Endpoint, remoteAddr.String())
+		}
+		return resolvedNodes
+	}
+
+	resolvedNodes := resolveNodes()
+
+	if len(resolvedNodes) == 0 {
+		logger.Info("No exit nodes available yet, waiting for nodes to be added")
+	} else {
+		// Send initial hole punch to all exit nodes
+		for _, node := range resolvedNodes {
+			if err := m.sendHolePunch(node.remoteAddr, node.publicKey); err != nil {
+				logger.Warn("Failed to send initial hole punch to %s: %v", node.endpointName, err)
+			}
+		}
+	}
+
+	// Start with minimum interval
+	m.mu.Lock()
+	m.sendHolepunchInterval = m.sendHolepunchIntervalMin
+	m.mu.Unlock()
+
+	ticker := time.NewTicker(m.sendHolepunchInterval)
+	defer ticker.Stop()
+
+	for {
+		select {
+		case <-m.stopChan:
+			logger.Debug("Hole punch stopped by signal")
+			return
+		case <-m.updateChan:
+			// Re-resolve exit nodes when update is signaled
+			logger.Info("Refreshing exit nodes for hole punching")
+			resolvedNodes = resolveNodes()
+			if len(resolvedNodes) == 0 {
+				logger.Warn("No exit nodes available after refresh")
+			} else {
+				logger.Info("Updated resolved nodes count: %d", len(resolvedNodes))
+			}
+			// Reset interval to minimum on update
+			m.mu.Lock()
+			m.sendHolepunchInterval = m.sendHolepunchIntervalMin
+			m.mu.Unlock()
+			ticker.Reset(m.sendHolepunchInterval)
+			// Send immediate hole punch to newly resolved nodes
+			for _, node := range resolvedNodes {
+				if err := m.sendHolePunch(node.remoteAddr, node.publicKey); err != nil {
+					logger.Debug("Failed to send hole punch to %s: %v", node.endpointName, err)
+				}
+			}
+		case <-ticker.C:
+			// Send hole punch to all exit nodes (if any are available)
+			if len(resolvedNodes) > 0 {
+				for _, node := range resolvedNodes {
+					if err := m.sendHolePunch(node.remoteAddr, node.publicKey); err != nil {
+						logger.Debug("Failed to send hole punch to %s: %v", node.endpointName, err)
+					}
+				}
+				// Exponential backoff: double the interval up to max
+				m.mu.Lock()
+				newInterval := m.sendHolepunchInterval * 2
+				if newInterval > m.sendHolepunchIntervalMax {
+					newInterval = m.sendHolepunchIntervalMax
+				}
+				if newInterval != m.sendHolepunchInterval {
+					m.sendHolepunchInterval = newInterval
+					ticker.Reset(m.sendHolepunchInterval)
+					logger.Debug("Increased hole punch interval to %v", m.sendHolepunchInterval)
+				}
+				m.mu.Unlock()
+			}
+		}
+	}
+}
+
+// sendHolePunch sends an encrypted hole punch packet using the shared bind
+func (m *Manager) sendHolePunch(remoteAddr *net.UDPAddr, serverPubKey string) error {
+	m.mu.Lock()
+	token := m.token
+	ID := m.ID
+	m.mu.Unlock()
+
+	if serverPubKey == "" || token == "" {
+		return fmt.Errorf("server public key or OLM token is empty")
+	}
+
+	var payload interface{}
+	if m.clientType == "newt" {
+		payload = struct {
+			ID        string `json:"newtId"`
+			Token     string `json:"token"`
+			PublicKey string `json:"publicKey"`
+		}{
+			ID:        ID,
+			Token:     token,
+			PublicKey: m.publicKey,
+		}
+	} else {
+		payload = struct {
+			ID        string `json:"olmId"`
+			Token     string `json:"token"`
+			PublicKey string `json:"publicKey"`
+		}{
+			ID:        ID,
+			Token:     token,
+			PublicKey: m.publicKey,
+		}
+	}
+
+	// Convert payload to JSON
+	payloadBytes, err := json.Marshal(payload)
+	if err != nil {
+		return fmt.Errorf("failed to marshal payload: %w", err)
+	}
+
+	// Encrypt the payload using the server's WireGuard public key
+	encryptedPayload, err := encryptPayload(payloadBytes, serverPubKey)
+	if err != nil {
+		return fmt.Errorf("failed to encrypt payload: %w", err)
+	}
+
+	jsonData, err := json.Marshal(encryptedPayload)
+	if err != nil {
+		return fmt.Errorf("failed to marshal encrypted payload: %w", err)
+	}
+
+	_, err = m.sharedBind.WriteToUDP(jsonData, remoteAddr)
+	if err != nil {
+		return fmt.Errorf("failed to write to UDP: %w", err)
+	}
+
+	logger.Debug("Sent UDP hole punch to %s: %s", remoteAddr.String(), string(jsonData))
+
+	return nil
+}
+
+// encryptPayload encrypts the payload using ChaCha20-Poly1305 AEAD with X25519 key exchange
+func encryptPayload(payload []byte, serverPublicKey string) (interface{}, error) {
+	// Generate an ephemeral keypair for this message
+	ephemeralPrivateKey, err := wgtypes.GeneratePrivateKey()
+	if err != nil {
+		return nil, fmt.Errorf("failed to generate ephemeral private key: %v", err)
+	}
+	ephemeralPublicKey := ephemeralPrivateKey.PublicKey()
+
+	// Parse the server's public key
+	serverPubKey, err := wgtypes.ParseKey(serverPublicKey)
+	if err != nil {
+		return nil, fmt.Errorf("failed to parse server public key: %v", err)
+	}
+
+	// Use X25519 for key exchange
+	var ephPrivKeyFixed [32]byte
+	copy(ephPrivKeyFixed[:], ephemeralPrivateKey[:])
+
+	// Perform X25519 key exchange
+	sharedSecret, err := curve25519.X25519(ephPrivKeyFixed[:], serverPubKey[:])
+	if err != nil {
+		return nil, fmt.Errorf("failed to perform X25519 key exchange: %v", err)
+	}
+
+	// Create an AEAD cipher using the shared secret
+	aead, err := chacha20poly1305.New(sharedSecret)
+	if err != nil {
+		return nil, fmt.Errorf("failed to create AEAD cipher: %v", err)
+	}
+
+	// Generate a random nonce
+	nonce := make([]byte, aead.NonceSize())
+	if _, err := mrand.Read(nonce); err != nil {
+		return nil, fmt.Errorf("failed to generate nonce: %v", err)
+	}
+
+	// Encrypt the payload
+	ciphertext := aead.Seal(nil, nonce, payload, nil)
+
+	// Prepare the final encrypted message
+	encryptedMsg := struct {
+		EphemeralPublicKey string `json:"ephemeralPublicKey"`
+		Nonce              []byte `json:"nonce"`
+		Ciphertext         []byte `json:"ciphertext"`
+	}{
+		EphemeralPublicKey: ephemeralPublicKey.String(),
+		Nonce:              nonce,
+		Ciphertext:         ciphertext,
+	}
+
+	return encryptedMsg, nil
+}

holepunch/tester.go

@@ -0,0 +1,424 @@
+package holepunch
+
+import (
+	"crypto/rand"
+	"fmt"
+	"net"
+	"net/netip"
+	"sync"
+	"time"
+
+	"github.com/fosrl/newt/bind"
+	"github.com/fosrl/newt/logger"
+	"github.com/fosrl/newt/util"
+)
+
+// TestResult represents the result of a connection test
+type TestResult struct {
+	// Success indicates whether the test was successful
+	Success bool
+	// RTT is the round-trip time of the test packet
+	RTT time.Duration
+	// Endpoint is the endpoint that was tested
+	Endpoint string
+	// Error contains any error that occurred during the test
+	Error error
+}
+
+// TestConnectionOptions configures the connection test
+type TestConnectionOptions struct {
+	// Timeout is how long to wait for a response (default: 5 seconds)
+	Timeout time.Duration
+	// Retries is the number of times to retry on failure (default: 0)
+	Retries int
+}
+
+// DefaultTestOptions returns the default test options
+func DefaultTestOptions() TestConnectionOptions {
+	return TestConnectionOptions{
+		Timeout: 5 * time.Second,
+		Retries: 0,
+	}
+}
+
+// cachedAddr holds a cached resolved UDP address
+type cachedAddr struct {
+	addr       *net.UDPAddr
+	resolvedAt time.Time
+}
+
+// HolepunchTester monitors holepunch connectivity using magic packets
+type HolepunchTester struct {
+	sharedBind *bind.SharedBind
+	publicDNS  []string
+	mu         sync.RWMutex
+	running    bool
+	stopChan   chan struct{}
+
+	// Pending requests waiting for responses (key: echo data as string)
+	pendingRequests sync.Map // map[string]*pendingRequest
+
+	// Callback when connection status changes
+	callback HolepunchStatusCallback
+
+	// Address cache to avoid repeated DNS/UDP resolution
+	addrCache    map[string]*cachedAddr
+	addrCacheMu  sync.RWMutex
+	addrCacheTTL time.Duration // How long cached addresses are valid
+}
+
+// HolepunchStatus represents the status of a holepunch connection
+type HolepunchStatus struct {
+	Endpoint  string
+	Connected bool
+	RTT       time.Duration
+}
+
+// HolepunchStatusCallback is called when holepunch status changes
+type HolepunchStatusCallback func(status HolepunchStatus)
+
+// pendingRequest tracks a pending test request
+type pendingRequest struct {
+	endpoint  string
+	sentAt    time.Time
+	replyChan chan time.Duration
+}
+
+// NewHolepunchTester creates a new holepunch tester using the given SharedBind
+func NewHolepunchTester(sharedBind *bind.SharedBind, publicDNS []string) *HolepunchTester {
+	return &HolepunchTester{
+		sharedBind:   sharedBind,
+		publicDNS:    publicDNS,
+		addrCache:    make(map[string]*cachedAddr),
+		addrCacheTTL: 5 * time.Minute, // Cache addresses for 5 minutes
+	}
+}
+
+// SetPublicDNS replaces the list of DNS servers used to resolve peer endpoints.
+// The servers must be in "host:port" format (e.g. "8.8.8.8:53").
+func (t *HolepunchTester) SetPublicDNS(servers []string) {
+	t.mu.Lock()
+	defer t.mu.Unlock()
+	t.publicDNS = servers
+	// Invalidate the address cache so endpoints are re-resolved with the new DNS.
+	t.addrCacheMu.Lock()
+	t.addrCache = make(map[string]*cachedAddr)
+	t.addrCacheMu.Unlock()
+}
+
+// SetCallback sets the callback for connection status changes
+func (t *HolepunchTester) SetCallback(callback HolepunchStatusCallback) {
+	t.mu.Lock()
+	defer t.mu.Unlock()
+	t.callback = callback
+}
+
+// Start begins listening for magic packet responses
+func (t *HolepunchTester) Start() error {
+	t.mu.Lock()
+	defer t.mu.Unlock()
+
+	if t.running {
+		return fmt.Errorf("tester already running")
+	}
+
+	if t.sharedBind == nil {
+		return fmt.Errorf("sharedBind is nil")
+	}
+
+	t.running = true
+	t.stopChan = make(chan struct{})
+
+	// Register our callback with the SharedBind to receive magic responses
+	t.sharedBind.SetMagicResponseCallback(t.handleResponse)
+
+	logger.Debug("HolepunchTester started")
+	return nil
+}
+
+// Stop stops the tester
+func (t *HolepunchTester) Stop() {
+	t.mu.Lock()
+	defer t.mu.Unlock()
+
+	if !t.running {
+		return
+	}
+
+	t.running = false
+	close(t.stopChan)
+
+	// Clear the callback
+	if t.sharedBind != nil {
+		t.sharedBind.SetMagicResponseCallback(nil)
+	}
+
+	// Cancel all pending requests
+	t.pendingRequests.Range(func(key, value interface{}) bool {
+		if req, ok := value.(*pendingRequest); ok {
+			close(req.replyChan)
+		}
+		t.pendingRequests.Delete(key)
+		return true
+	})
+
+	// Clear address cache
+	t.addrCacheMu.Lock()
+	t.addrCache = make(map[string]*cachedAddr)
+	t.addrCacheMu.Unlock()
+
+	logger.Debug("HolepunchTester stopped")
+}
+
+// resolveEndpoint resolves an endpoint to a UDP address, using cache when possible
+func (t *HolepunchTester) resolveEndpoint(endpoint string) (*net.UDPAddr, error) {
+	// Check cache first
+	t.addrCacheMu.RLock()
+	cached, ok := t.addrCache[endpoint]
+	ttl := t.addrCacheTTL
+	t.addrCacheMu.RUnlock()
+
+	if ok && time.Since(cached.resolvedAt) < ttl {
+		return cached.addr, nil
+	}
+
+	// Resolve the endpoint
+	var host string
+	var err error
+	if len(t.publicDNS) > 0 {
+		host, err = util.ResolveDomainUpstream(endpoint, t.publicDNS)
+	} else {
+		host, err = util.ResolveDomain(endpoint)
+	}
+	if err != nil {
+		host = endpoint
+	}
+
+	_, _, err = net.SplitHostPort(host)
+	if err != nil {
+		host = net.JoinHostPort(host, "21820")
+	}
+
+	remoteAddr, err := net.ResolveUDPAddr("udp", host)
+	if err != nil {
+		return nil, fmt.Errorf("failed to resolve UDP address %s: %w", host, err)
+	}
+
+	// Cache the result
+	t.addrCacheMu.Lock()
+	t.addrCache[endpoint] = &cachedAddr{
+		addr:       remoteAddr,
+		resolvedAt: time.Now(),
+	}
+	t.addrCacheMu.Unlock()
+
+	return remoteAddr, nil
+}
+
+// InvalidateCache removes a specific endpoint from the address cache
+func (t *HolepunchTester) InvalidateCache(endpoint string) {
+	t.addrCacheMu.Lock()
+	delete(t.addrCache, endpoint)
+	t.addrCacheMu.Unlock()
+}
+
+// ClearCache clears all cached addresses
+func (t *HolepunchTester) ClearCache() {
+	t.addrCacheMu.Lock()
+	t.addrCache = make(map[string]*cachedAddr)
+	t.addrCacheMu.Unlock()
+}
+
+// handleResponse is called by SharedBind when a magic response is received
+func (t *HolepunchTester) handleResponse(addr netip.AddrPort, echoData []byte) {
+	// logger.Debug("Received magic response from %s", addr.String())
+	key := string(echoData)
+
+	value, ok := t.pendingRequests.LoadAndDelete(key)
+	if !ok {
+		// No matching request found
+		logger.Debug("No pending request found for magic response from %s", addr.String())
+		return
+	}
+
+	req := value.(*pendingRequest)
+	rtt := time.Since(req.sentAt)
+	// logger.Debug("Magic response matched pending request for %s (RTT: %v)", req.endpoint, rtt)
+
+	// Send RTT to the waiting goroutine (non-blocking)
+	select {
+	case req.replyChan <- rtt:
+	default:
+	}
+}
+
+// TestEndpoint sends a magic test packet to the endpoint and waits for a response.
+// This uses the SharedBind so packets come from the same source port as WireGuard.
+func (t *HolepunchTester) TestEndpoint(endpoint string, timeout time.Duration) TestResult {
+	result := TestResult{
+		Endpoint: endpoint,
+	}
+
+	t.mu.RLock()
+	running := t.running
+	sharedBind := t.sharedBind
+	t.mu.RUnlock()
+
+	if !running {
+		result.Error = fmt.Errorf("tester not running")
+		return result
+	}
+
+	if sharedBind == nil || sharedBind.IsClosed() {
+		result.Error = fmt.Errorf("sharedBind is nil or closed")
+		return result
+	}
+
+	// Resolve the endpoint (using cache)
+	remoteAddr, err := t.resolveEndpoint(endpoint)
+	if err != nil {
+		result.Error = err
+		return result
+	}
+
+	// Generate random data for the test packet
+	randomData := make([]byte, bind.MagicPacketDataLen)
+	if _, err := rand.Read(randomData); err != nil {
+		result.Error = fmt.Errorf("failed to generate random data: %w", err)
+		return result
+	}
+
+	// Create a pending request
+	req := &pendingRequest{
+		endpoint:  endpoint,
+		sentAt:    time.Now(),
+		replyChan: make(chan time.Duration, 1),
+	}
+
+	key := string(randomData)
+	t.pendingRequests.Store(key, req)
+
+	// Build the test request packet
+	request := make([]byte, bind.MagicTestRequestLen)
+	copy(request, bind.MagicTestRequest)
+	copy(request[len(bind.MagicTestRequest):], randomData)
+
+	// Send the test packet
+	_, err = sharedBind.WriteToUDP(request, remoteAddr)
+	if err != nil {
+		t.pendingRequests.Delete(key)
+		result.Error = fmt.Errorf("failed to send test packet: %w", err)
+		return result
+	}
+
+	// Wait for response with timeout
+	select {
+	case rtt, ok := <-req.replyChan:
+		if ok {
+			result.Success = true
+			result.RTT = rtt
+		} else {
+			result.Error = fmt.Errorf("request cancelled")
+		}
+	case <-time.After(timeout):
+		t.pendingRequests.Delete(key)
+		result.Error = fmt.Errorf("timeout waiting for response")
+	}
+
+	return result
+}
+
+// TestConnectionWithBind sends a magic test packet using an existing SharedBind.
+// This is useful when you want to test the connection through the same socket
+// that WireGuard is using, which tests the actual hole-punched path.
+func TestConnectionWithBind(sharedBind *bind.SharedBind, endpoint string, opts *TestConnectionOptions) TestResult {
+	if opts == nil {
+		defaultOpts := DefaultTestOptions()
+		opts = &defaultOpts
+	}
+
+	result := TestResult{
+		Endpoint: endpoint,
+	}
+
+	if sharedBind == nil {
+		result.Error = fmt.Errorf("sharedBind is nil")
+		return result
+	}
+
+	if sharedBind.IsClosed() {
+		result.Error = fmt.Errorf("sharedBind is closed")
+		return result
+	}
+
+	// Resolve the endpoint
+	host, err := util.ResolveDomain(endpoint)
+	if err != nil {
+		host = endpoint
+	}
+
+	_, _, err = net.SplitHostPort(host)
+	if err != nil {
+		host = net.JoinHostPort(host, "21820")
+	}
+
+	remoteAddr, err := net.ResolveUDPAddr("udp", host)
+	if err != nil {
+		result.Error = fmt.Errorf("failed to resolve UDP address %s: %w", host, err)
+		return result
+	}
+
+	// Generate random data for the test packet
+	randomData := make([]byte, bind.MagicPacketDataLen)
+	if _, err := rand.Read(randomData); err != nil {
+		result.Error = fmt.Errorf("failed to generate random data: %w", err)
+		return result
+	}
+
+	// Build the test request packet
+	request := make([]byte, bind.MagicTestRequestLen)
+	copy(request, bind.MagicTestRequest)
+	copy(request[len(bind.MagicTestRequest):], randomData)
+
+	// Get the underlying UDP connection to set read deadline and read response
+	udpConn := sharedBind.GetUDPConn()
+	if udpConn == nil {
+		result.Error = fmt.Errorf("could not get UDP connection from SharedBind")
+		return result
+	}
+
+	attempts := opts.Retries + 1
+	for attempt := 0; attempt < attempts; attempt++ {
+		if attempt > 0 {
+			logger.Debug("Retrying connection test to %s (attempt %d/%d)", endpoint, attempt+1, attempts)
+		}
+
+		// Note: We can't easily set a read deadline on the shared connection
+		// without affecting WireGuard, so we use a goroutine with timeout instead
+		startTime := time.Now()
+
+		// Send the test packet through the shared bind
+		_, err = sharedBind.WriteToUDP(request, remoteAddr)
+		if err != nil {
+			result.Error = fmt.Errorf("failed to send test packet: %w", err)
+			if attempt < attempts-1 {
+				continue
+			}
+			return result
+		}
+
+		// For shared bind test, we send the packet but can't easily wait for
+		// response without interfering with WireGuard's receive loop.
+		// The response will be handled by SharedBind automatically.
+		// We consider the test successful if the send succeeded.
+		// For a full round-trip test, use TestConnection() with a separate socket.
+
+		result.RTT = time.Since(startTime)
+		result.Success = true
+		result.Error = nil
+		logger.Debug("Test packet sent to %s via SharedBind", endpoint)
+		return result
+	}
+
+	return result
+}

internal/state/telemetry_view.go

@@ -0,0 +1,80 @@
+package state
+
+import (
+	"sync"
+	"sync/atomic"
+	"time"
+
+	"github.com/fosrl/newt/internal/telemetry"
+)
+
+// TelemetryView is a minimal, thread-safe implementation to feed observables.
+// Since one Newt process represents one site, we expose a single logical site.
+// site_id is a resource attribute, so we do not emit per-site labels here.
+type TelemetryView struct {
+	online     atomic.Bool
+	lastHBUnix atomic.Int64 // unix seconds
+	// per-tunnel sessions
+	sessMu   sync.RWMutex
+	sessions map[string]*atomic.Int64
+}
+
+var (
+	globalView atomic.Pointer[TelemetryView]
+)
+
+// Global returns a singleton TelemetryView.
+func Global() *TelemetryView {
+	if v := globalView.Load(); v != nil { return v }
+	v := &TelemetryView{ sessions: make(map[string]*atomic.Int64) }
+	globalView.Store(v)
+	telemetry.RegisterStateView(v)
+	return v
+}
+
+// Instrumentation helpers
+func (v *TelemetryView) IncSessions(tunnelID string) {
+	v.sessMu.Lock(); defer v.sessMu.Unlock()
+	c := v.sessions[tunnelID]
+	if c == nil { c = &atomic.Int64{}; v.sessions[tunnelID] = c }
+	c.Add(1)
+}
+func (v *TelemetryView) DecSessions(tunnelID string) {
+	v.sessMu.Lock(); defer v.sessMu.Unlock()
+	if c := v.sessions[tunnelID]; c != nil {
+		c.Add(-1)
+		if c.Load() <= 0 { delete(v.sessions, tunnelID) }
+	}
+}
+func (v *TelemetryView) ClearTunnel(tunnelID string) {
+	v.sessMu.Lock(); defer v.sessMu.Unlock()
+	delete(v.sessions, tunnelID)
+}
+func (v *TelemetryView) SetOnline(b bool) { v.online.Store(b) }
+func (v *TelemetryView) TouchHeartbeat() { v.lastHBUnix.Store(time.Now().Unix()) }
+
+// --- telemetry.StateView interface ---
+
+func (v *TelemetryView) ListSites() []string { return []string{"self"} }
+func (v *TelemetryView) Online(_ string) (bool, bool) { return v.online.Load(), true }
+func (v *TelemetryView) LastHeartbeat(_ string) (time.Time, bool) {
+	sec := v.lastHBUnix.Load()
+	if sec == 0 { return time.Time{}, false }
+	return time.Unix(sec, 0), true
+}
+func (v *TelemetryView) ActiveSessions(_ string) (int64, bool) {
+	// aggregated sessions (not used for per-tunnel gauge)
+	v.sessMu.RLock(); defer v.sessMu.RUnlock()
+	var sum int64
+	for _, c := range v.sessions { if c != nil { sum += c.Load() } }
+	return sum, true
+}
+
+// Extended accessor used by telemetry callback to publish per-tunnel samples.
+func (v *TelemetryView) SessionsByTunnel() map[string]int64 {
+	v.sessMu.RLock(); defer v.sessMu.RUnlock()
+	out := make(map[string]int64, len(v.sessions))
+	for id, c := range v.sessions { if c != nil && c.Load() > 0 { out[id] = c.Load() } }
+	return out
+}
+

internal/telemetry/constants.go

@@ -0,0 +1,19 @@
+package telemetry
+
+// Protocol labels (low-cardinality)
+const (
+	ProtocolTCP = "tcp"
+	ProtocolUDP = "udp"
+)
+
+// Reconnect reason bins (fixed, low-cardinality)
+const (
+	ReasonServerRequest  = "server_request"
+	ReasonTimeout        = "timeout"
+	ReasonPeerClose      = "peer_close"
+	ReasonNetworkChange  = "network_change"
+	ReasonAuthError      = "auth_error"
+	ReasonHandshakeError = "handshake_error"
+	ReasonConfigChange   = "config_change"
+	ReasonError          = "error"
+)

internal/telemetry/constants_test.go

@@ -0,0 +1,32 @@
+package telemetry
+
+import "testing"
+
+func TestAllowedConstants(t *testing.T) {
+	allowedReasons := map[string]struct{}{
+		ReasonServerRequest:  {},
+		ReasonTimeout:        {},
+		ReasonPeerClose:      {},
+		ReasonNetworkChange:  {},
+		ReasonAuthError:      {},
+		ReasonHandshakeError: {},
+		ReasonConfigChange:   {},
+		ReasonError:          {},
+	}
+	for k := range allowedReasons {
+		if k == "" {
+			t.Fatalf("empty reason constant")
+		}
+	}
+
+	allowedProtocols := map[string]struct{}{
+		ProtocolTCP: {},
+		ProtocolUDP: {},
+	}
+	for k := range allowedProtocols {
+		if k == "" {
+			t.Fatalf("empty protocol constant")
+		}
+	}
+}
+

internal/telemetry/metrics.go

@@ -0,0 +1,542 @@
+package telemetry
+
+import (
+	"context"
+	"sync"
+	"sync/atomic"
+	"time"
+
+	"go.opentelemetry.io/otel"
+	"go.opentelemetry.io/otel/attribute"
+	"go.opentelemetry.io/otel/metric"
+)
+
+// Instruments and helpers for Newt metrics following the naming, units, and
+// low-cardinality label guidance from the issue description.
+//
+// Counters end with _total, durations are in seconds, sizes in bytes.
+// Only low-cardinality stable labels are supported: tunnel_id,
+// transport, direction, result, reason, error_type.
+var (
+	initOnce sync.Once
+
+	meter metric.Meter
+
+	// Site / Registration
+	mSiteRegistrations metric.Int64Counter
+	mSiteOnline        metric.Int64ObservableGauge
+	mSiteLastHeartbeat metric.Float64ObservableGauge
+
+	// Tunnel / Sessions
+	mTunnelSessions metric.Int64ObservableGauge
+	mTunnelBytes    metric.Int64Counter
+	mTunnelLatency  metric.Float64Histogram
+	mReconnects     metric.Int64Counter
+
+	// Connection / NAT
+	mConnAttempts metric.Int64Counter
+	mConnErrors   metric.Int64Counter
+
+	// Config/Restart
+	mConfigReloads     metric.Int64Counter
+	mConfigApply       metric.Float64Histogram
+	mCertRotationTotal metric.Int64Counter
+	mProcessStartTime  metric.Float64ObservableGauge
+
+	// Build info
+	mBuildInfo metric.Int64ObservableGauge
+
+	// WebSocket
+	mWSConnectLatency   metric.Float64Histogram
+	mWSMessages         metric.Int64Counter
+	mWSDisconnects      metric.Int64Counter
+	mWSKeepaliveFailure metric.Int64Counter
+	mWSSessionDuration  metric.Float64Histogram
+	mWSConnected        metric.Int64ObservableGauge
+	mWSReconnects       metric.Int64Counter
+
+	// Proxy
+	mProxyActiveConns      metric.Int64ObservableGauge
+	mProxyBufferBytes      metric.Int64ObservableGauge
+	mProxyAsyncBacklogByte metric.Int64ObservableGauge
+	mProxyDropsTotal       metric.Int64Counter
+	mProxyAcceptsTotal     metric.Int64Counter
+	mProxyConnDuration     metric.Float64Histogram
+	mProxyConnectionsTotal metric.Int64Counter
+
+	buildVersion     string
+	buildCommit      string
+	processStartUnix = float64(time.Now().UnixNano()) / 1e9
+	wsConnectedState atomic.Int64
+)
+
+// Proxy connection lifecycle events.
+const (
+	ProxyConnectionOpened = "opened"
+	ProxyConnectionClosed = "closed"
+)
+
+// attrsWithSite appends site/region labels only when explicitly enabled to keep
+// label cardinality low by default.
+func attrsWithSite(extra ...attribute.KeyValue) []attribute.KeyValue {
+	attrs := make([]attribute.KeyValue, len(extra))
+	copy(attrs, extra)
+	if ShouldIncludeSiteLabels() {
+		attrs = append(attrs, siteAttrs()...)
+	}
+	return attrs
+}
+
+func registerInstruments() error {
+	var err error
+	initOnce.Do(func() {
+		meter = otel.Meter("newt")
+		if e := registerSiteInstruments(); e != nil {
+			err = e
+			return
+		}
+		if e := registerTunnelInstruments(); e != nil {
+			err = e
+			return
+		}
+		if e := registerConnInstruments(); e != nil {
+			err = e
+			return
+		}
+		if e := registerConfigInstruments(); e != nil {
+			err = e
+			return
+		}
+		if e := registerBuildWSProxyInstruments(); e != nil {
+			err = e
+			return
+		}
+	})
+	return err
+}
+
+func registerSiteInstruments() error {
+	var err error
+	mSiteRegistrations, err = meter.Int64Counter("newt_site_registrations_total",
+		metric.WithDescription("Total site registration attempts"))
+	if err != nil {
+		return err
+	}
+	mSiteOnline, err = meter.Int64ObservableGauge("newt_site_online",
+		metric.WithDescription("Site online (0/1)"))
+	if err != nil {
+		return err
+	}
+	mSiteLastHeartbeat, err = meter.Float64ObservableGauge("newt_site_last_heartbeat_timestamp_seconds",
+		metric.WithDescription("Unix timestamp of the last site heartbeat"),
+		metric.WithUnit("s"))
+	if err != nil {
+		return err
+	}
+	return nil
+}
+
+func registerTunnelInstruments() error {
+	var err error
+	mTunnelSessions, err = meter.Int64ObservableGauge("newt_tunnel_sessions",
+		metric.WithDescription("Active tunnel sessions"))
+	if err != nil {
+		return err
+	}
+	mTunnelBytes, err = meter.Int64Counter("newt_tunnel_bytes_total",
+		metric.WithDescription("Tunnel bytes ingress/egress"),
+		metric.WithUnit("By"))
+	if err != nil {
+		return err
+	}
+	mTunnelLatency, err = meter.Float64Histogram("newt_tunnel_latency_seconds",
+		metric.WithDescription("Per-tunnel latency in seconds"),
+		metric.WithUnit("s"))
+	if err != nil {
+		return err
+	}
+	mReconnects, err = meter.Int64Counter("newt_tunnel_reconnects_total",
+		metric.WithDescription("Tunnel reconnect events"))
+	if err != nil {
+		return err
+	}
+	return nil
+}
+
+func registerConnInstruments() error {
+	var err error
+	mConnAttempts, err = meter.Int64Counter("newt_connection_attempts_total",
+		metric.WithDescription("Connection attempts"))
+	if err != nil {
+		return err
+	}
+	mConnErrors, err = meter.Int64Counter("newt_connection_errors_total",
+		metric.WithDescription("Connection errors by type"))
+	if err != nil {
+		return err
+	}
+	return nil
+}
+
+func registerConfigInstruments() error {
+	mConfigReloads, _ = meter.Int64Counter("newt_config_reloads_total",
+		metric.WithDescription("Configuration reloads"))
+	mConfigApply, _ = meter.Float64Histogram("newt_config_apply_seconds",
+		metric.WithDescription("Configuration apply duration in seconds"),
+		metric.WithUnit("s"))
+	mCertRotationTotal, _ = meter.Int64Counter("newt_cert_rotation_total",
+		metric.WithDescription("Certificate rotation events (success/failure)"))
+	mProcessStartTime, _ = meter.Float64ObservableGauge("process_start_time_seconds",
+		metric.WithDescription("Unix timestamp of the process start time"),
+		metric.WithUnit("s"))
+	if mProcessStartTime != nil {
+		if _, err := meter.RegisterCallback(func(ctx context.Context, o metric.Observer) error {
+			o.ObserveFloat64(mProcessStartTime, processStartUnix)
+			return nil
+		}, mProcessStartTime); err != nil {
+			otel.Handle(err)
+		}
+	}
+	return nil
+}
+
+func registerBuildWSProxyInstruments() error {
+	// Build info gauge (value 1 with version/commit attributes)
+	mBuildInfo, _ = meter.Int64ObservableGauge("newt_build_info",
+		metric.WithDescription("Newt build information (value is always 1)"))
+	// WebSocket
+	mWSConnectLatency, _ = meter.Float64Histogram("newt_websocket_connect_latency_seconds",
+		metric.WithDescription("WebSocket connect latency in seconds"),
+		metric.WithUnit("s"))
+	mWSMessages, _ = meter.Int64Counter("newt_websocket_messages_total",
+		metric.WithDescription("WebSocket messages by direction and type"))
+	mWSDisconnects, _ = meter.Int64Counter("newt_websocket_disconnects_total",
+		metric.WithDescription("WebSocket disconnects by reason/result"))
+	mWSKeepaliveFailure, _ = meter.Int64Counter("newt_websocket_keepalive_failures_total",
+		metric.WithDescription("WebSocket keepalive (ping/pong) failures"))
+	mWSSessionDuration, _ = meter.Float64Histogram("newt_websocket_session_duration_seconds",
+		metric.WithDescription("Duration of established WebSocket sessions"),
+		metric.WithUnit("s"))
+	mWSConnected, _ = meter.Int64ObservableGauge("newt_websocket_connected",
+		metric.WithDescription("WebSocket connection state (1=connected, 0=disconnected)"))
+	mWSReconnects, _ = meter.Int64Counter("newt_websocket_reconnects_total",
+		metric.WithDescription("WebSocket reconnect attempts by reason"))
+	// Proxy
+	mProxyActiveConns, _ = meter.Int64ObservableGauge("newt_proxy_active_connections",
+		metric.WithDescription("Proxy active connections per tunnel and protocol"))
+	mProxyBufferBytes, _ = meter.Int64ObservableGauge("newt_proxy_buffer_bytes",
+		metric.WithDescription("Proxy buffer bytes (may approximate async backlog)"),
+		metric.WithUnit("By"))
+	mProxyAsyncBacklogByte, _ = meter.Int64ObservableGauge("newt_proxy_async_backlog_bytes",
+		metric.WithDescription("Unflushed async byte backlog per tunnel and protocol"),
+		metric.WithUnit("By"))
+	mProxyDropsTotal, _ = meter.Int64Counter("newt_proxy_drops_total",
+		metric.WithDescription("Proxy drops due to write errors"))
+	mProxyAcceptsTotal, _ = meter.Int64Counter("newt_proxy_accept_total",
+		metric.WithDescription("Proxy connection accepts by protocol and result"))
+	mProxyConnDuration, _ = meter.Float64Histogram("newt_proxy_connection_duration_seconds",
+		metric.WithDescription("Duration of completed proxy connections"),
+		metric.WithUnit("s"))
+	mProxyConnectionsTotal, _ = meter.Int64Counter("newt_proxy_connections_total",
+		metric.WithDescription("Proxy connection lifecycle events by protocol"))
+	// Register a default callback for build info if version/commit set
+	reg, e := meter.RegisterCallback(func(ctx context.Context, o metric.Observer) error {
+		if buildVersion == "" && buildCommit == "" {
+			return nil
+		}
+		attrs := []attribute.KeyValue{}
+		if buildVersion != "" {
+			attrs = append(attrs, attribute.String("version", buildVersion))
+		}
+		if buildCommit != "" {
+			attrs = append(attrs, attribute.String("commit", buildCommit))
+		}
+		if ShouldIncludeSiteLabels() {
+			attrs = append(attrs, siteAttrs()...)
+		}
+		o.ObserveInt64(mBuildInfo, 1, metric.WithAttributes(attrs...))
+		return nil
+	}, mBuildInfo)
+	if e != nil {
+		otel.Handle(e)
+	} else {
+		// Provide a functional stopper that unregisters the callback
+		obsStopper = func() { _ = reg.Unregister() }
+	}
+	if mWSConnected != nil {
+		if regConn, err := meter.RegisterCallback(func(ctx context.Context, o metric.Observer) error {
+			val := wsConnectedState.Load()
+			o.ObserveInt64(mWSConnected, val, metric.WithAttributes(attrsWithSite()...))
+			return nil
+		}, mWSConnected); err != nil {
+			otel.Handle(err)
+		} else {
+			wsConnStopper = func() { _ = regConn.Unregister() }
+		}
+	}
+	return nil
+}
+
+// Observable registration: Newt can register a callback to report gauges.
+// Call SetObservableCallback once to start observing online status, last
+// heartbeat seconds, and active sessions.
+
+var (
+	obsOnce       sync.Once
+	obsStopper    func()
+	proxyObsOnce  sync.Once
+	proxyStopper  func()
+	wsConnStopper func()
+)
+
+// SetObservableCallback registers a single callback that will be invoked
+// on collection. Use the provided observer to emit values for the observable
+// gauges defined here.
+//
+// Example inside your code (where you have access to current state):
+//
+//	telemetry.SetObservableCallback(func(ctx context.Context, o metric.Observer) error {
+//	    o.ObserveInt64(mSiteOnline, 1)
+//	    o.ObserveFloat64(mSiteLastHeartbeat, float64(lastHB.Unix()))
+//	    o.ObserveInt64(mTunnelSessions, int64(len(activeSessions)))
+//	    return nil
+//	})
+func SetObservableCallback(cb func(context.Context, metric.Observer) error) {
+	obsOnce.Do(func() {
+		reg, e := meter.RegisterCallback(cb, mSiteOnline, mSiteLastHeartbeat, mTunnelSessions)
+		if e != nil {
+			otel.Handle(e)
+			obsStopper = func() {
+				// no-op: registration failed; keep stopper callable
+			}
+			return
+		}
+		// Provide a functional stopper mirroring proxy/build-info behavior
+		obsStopper = func() { _ = reg.Unregister() }
+	})
+}
+
+// SetProxyObservableCallback registers a callback to observe proxy gauges.
+func SetProxyObservableCallback(cb func(context.Context, metric.Observer) error) {
+	proxyObsOnce.Do(func() {
+		reg, e := meter.RegisterCallback(cb, mProxyActiveConns, mProxyBufferBytes, mProxyAsyncBacklogByte)
+		if e != nil {
+			otel.Handle(e)
+			proxyStopper = func() {
+				// no-op: registration failed; keep stopper callable
+			}
+			return
+		}
+		// Provide a functional stopper to unregister later if needed
+		proxyStopper = func() { _ = reg.Unregister() }
+	})
+}
+
+// Build info registration
+func RegisterBuildInfo(version, commit string) {
+	buildVersion = version
+	buildCommit = commit
+}
+
+// Config reloads
+func IncConfigReload(ctx context.Context, result string) {
+	mConfigReloads.Add(ctx, 1, metric.WithAttributes(attrsWithSite(
+		attribute.String("result", result),
+	)...))
+}
+
+// Helpers for counters/histograms
+
+func IncSiteRegistration(ctx context.Context, result string) {
+	attrs := []attribute.KeyValue{
+		attribute.String("result", result),
+	}
+	mSiteRegistrations.Add(ctx, 1, metric.WithAttributes(attrsWithSite(attrs...)...))
+}
+
+func AddTunnelBytes(ctx context.Context, tunnelID, direction string, n int64) {
+	attrs := []attribute.KeyValue{
+		attribute.String("direction", direction),
+	}
+	if ShouldIncludeTunnelID() && tunnelID != "" {
+		attrs = append(attrs, attribute.String("tunnel_id", tunnelID))
+	}
+	mTunnelBytes.Add(ctx, n, metric.WithAttributes(attrsWithSite(attrs...)...))
+}
+
+// AddTunnelBytesSet adds bytes using a pre-built attribute.Set to avoid per-call allocations.
+func AddTunnelBytesSet(ctx context.Context, n int64, attrs attribute.Set) {
+	mTunnelBytes.Add(ctx, n, metric.WithAttributeSet(attrs))
+}
+
+// --- WebSocket helpers ---
+
+func ObserveWSConnectLatency(ctx context.Context, seconds float64, result, errorType string) {
+	attrs := []attribute.KeyValue{
+		attribute.String("transport", "websocket"),
+		attribute.String("result", result),
+	}
+	if errorType != "" {
+		attrs = append(attrs, attribute.String("error_type", errorType))
+	}
+	mWSConnectLatency.Record(ctx, seconds, metric.WithAttributes(attrsWithSite(attrs...)...))
+}
+
+func IncWSMessage(ctx context.Context, direction, msgType string) {
+	mWSMessages.Add(ctx, 1, metric.WithAttributes(attrsWithSite(
+		attribute.String("direction", direction),
+		attribute.String("msg_type", msgType),
+	)...))
+}
+
+func IncWSDisconnect(ctx context.Context, reason, result string) {
+	mWSDisconnects.Add(ctx, 1, metric.WithAttributes(attrsWithSite(
+		attribute.String("reason", reason),
+		attribute.String("result", result),
+	)...))
+}
+
+func IncWSKeepaliveFailure(ctx context.Context, reason string) {
+	mWSKeepaliveFailure.Add(ctx, 1, metric.WithAttributes(attrsWithSite(
+		attribute.String("reason", reason),
+	)...))
+}
+
+// SetWSConnectionState updates the backing gauge for the WebSocket connected state.
+func SetWSConnectionState(connected bool) {
+	if connected {
+		wsConnectedState.Store(1)
+	} else {
+		wsConnectedState.Store(0)
+	}
+}
+
+// IncWSReconnect increments the WebSocket reconnect counter with a bounded reason label.
+func IncWSReconnect(ctx context.Context, reason string) {
+	if reason == "" {
+		reason = "unknown"
+	}
+	mWSReconnects.Add(ctx, 1, metric.WithAttributes(attrsWithSite(
+		attribute.String("reason", reason),
+	)...))
+}
+
+func ObserveWSSessionDuration(ctx context.Context, seconds float64, result string) {
+	mWSSessionDuration.Record(ctx, seconds, metric.WithAttributes(attrsWithSite(
+		attribute.String("result", result),
+	)...))
+}
+
+// --- Proxy helpers ---
+
+func ObserveProxyActiveConnsObs(o metric.Observer, value int64, attrs []attribute.KeyValue) {
+	o.ObserveInt64(mProxyActiveConns, value, metric.WithAttributes(attrs...))
+}
+
+func ObserveProxyBufferBytesObs(o metric.Observer, value int64, attrs []attribute.KeyValue) {
+	o.ObserveInt64(mProxyBufferBytes, value, metric.WithAttributes(attrs...))
+}
+
+func ObserveProxyAsyncBacklogObs(o metric.Observer, value int64, attrs []attribute.KeyValue) {
+	o.ObserveInt64(mProxyAsyncBacklogByte, value, metric.WithAttributes(attrs...))
+}
+
+func IncProxyDrops(ctx context.Context, tunnelID, protocol string) {
+	attrs := []attribute.KeyValue{
+		attribute.String("protocol", protocol),
+	}
+	if ShouldIncludeTunnelID() && tunnelID != "" {
+		attrs = append(attrs, attribute.String("tunnel_id", tunnelID))
+	}
+	mProxyDropsTotal.Add(ctx, 1, metric.WithAttributes(attrsWithSite(attrs...)...))
+}
+
+func IncProxyAccept(ctx context.Context, tunnelID, protocol, result, reason string) {
+	attrs := []attribute.KeyValue{
+		attribute.String("protocol", protocol),
+		attribute.String("result", result),
+	}
+	if reason != "" {
+		attrs = append(attrs, attribute.String("reason", reason))
+	}
+	if ShouldIncludeTunnelID() && tunnelID != "" {
+		attrs = append(attrs, attribute.String("tunnel_id", tunnelID))
+	}
+	mProxyAcceptsTotal.Add(ctx, 1, metric.WithAttributes(attrsWithSite(attrs...)...))
+}
+
+func ObserveProxyConnectionDuration(ctx context.Context, tunnelID, protocol, result string, seconds float64) {
+	attrs := []attribute.KeyValue{
+		attribute.String("protocol", protocol),
+		attribute.String("result", result),
+	}
+	if ShouldIncludeTunnelID() && tunnelID != "" {
+		attrs = append(attrs, attribute.String("tunnel_id", tunnelID))
+	}
+	mProxyConnDuration.Record(ctx, seconds, metric.WithAttributes(attrsWithSite(attrs...)...))
+}
+
+// IncProxyConnectionEvent records proxy connection lifecycle events (opened/closed).
+func IncProxyConnectionEvent(ctx context.Context, tunnelID, protocol, event string) {
+	if event == "" {
+		event = "unknown"
+	}
+	attrs := []attribute.KeyValue{
+		attribute.String("protocol", protocol),
+		attribute.String("event", event),
+	}
+	if ShouldIncludeTunnelID() && tunnelID != "" {
+		attrs = append(attrs, attribute.String("tunnel_id", tunnelID))
+	}
+	mProxyConnectionsTotal.Add(ctx, 1, metric.WithAttributes(attrsWithSite(attrs...)...))
+}
+
+// --- Config/PKI helpers ---
+
+func ObserveConfigApply(ctx context.Context, phase, result string, seconds float64) {
+	mConfigApply.Record(ctx, seconds, metric.WithAttributes(attrsWithSite(
+		attribute.String("phase", phase),
+		attribute.String("result", result),
+	)...))
+}
+
+func IncCertRotation(ctx context.Context, result string) {
+	mCertRotationTotal.Add(ctx, 1, metric.WithAttributes(attrsWithSite(
+		attribute.String("result", result),
+	)...))
+}
+
+func ObserveTunnelLatency(ctx context.Context, tunnelID, transport string, seconds float64) {
+	attrs := []attribute.KeyValue{
+		attribute.String("transport", transport),
+	}
+	if ShouldIncludeTunnelID() && tunnelID != "" {
+		attrs = append(attrs, attribute.String("tunnel_id", tunnelID))
+	}
+	mTunnelLatency.Record(ctx, seconds, metric.WithAttributes(attrsWithSite(attrs...)...))
+}
+
+func IncReconnect(ctx context.Context, tunnelID, initiator, reason string) {
+	attrs := []attribute.KeyValue{
+		attribute.String("initiator", initiator),
+		attribute.String("reason", reason),
+	}
+	if ShouldIncludeTunnelID() && tunnelID != "" {
+		attrs = append(attrs, attribute.String("tunnel_id", tunnelID))
+	}
+	mReconnects.Add(ctx, 1, metric.WithAttributes(attrsWithSite(attrs...)...))
+}
+
+func IncConnAttempt(ctx context.Context, transport, result string) {
+	mConnAttempts.Add(ctx, 1, metric.WithAttributes(attrsWithSite(
+		attribute.String("transport", transport),
+		attribute.String("result", result),
+	)...))
+}
+
+func IncConnError(ctx context.Context, transport, typ string) {
+	mConnErrors.Add(ctx, 1, metric.WithAttributes(attrsWithSite(
+		attribute.String("transport", transport),
+		attribute.String("error_type", typ),
+	)...))
+}

internal/telemetry/metrics_test_helper.go

@@ -0,0 +1,59 @@
+package telemetry
+
+import (
+	"sync"
+	"time"
+)
+
+func resetMetricsForTest() {
+	initOnce = sync.Once{}
+	obsOnce = sync.Once{}
+	proxyObsOnce = sync.Once{}
+	obsStopper = nil
+	proxyStopper = nil
+	if wsConnStopper != nil {
+		wsConnStopper()
+	}
+	wsConnStopper = nil
+	meter = nil
+
+	mSiteRegistrations = nil
+	mSiteOnline = nil
+	mSiteLastHeartbeat = nil
+
+	mTunnelSessions = nil
+	mTunnelBytes = nil
+	mTunnelLatency = nil
+	mReconnects = nil
+
+	mConnAttempts = nil
+	mConnErrors = nil
+
+	mConfigReloads = nil
+	mConfigApply = nil
+	mCertRotationTotal = nil
+	mProcessStartTime = nil
+
+	mBuildInfo = nil
+
+	mWSConnectLatency = nil
+	mWSMessages = nil
+	mWSDisconnects = nil
+	mWSKeepaliveFailure = nil
+	mWSSessionDuration = nil
+	mWSConnected = nil
+	mWSReconnects = nil
+
+	mProxyActiveConns = nil
+	mProxyBufferBytes = nil
+	mProxyAsyncBacklogByte = nil
+	mProxyDropsTotal = nil
+	mProxyAcceptsTotal = nil
+	mProxyConnDuration = nil
+	mProxyConnectionsTotal = nil
+
+	processStartUnix = float64(time.Now().UnixNano()) / 1e9
+	wsConnectedState.Store(0)
+	includeTunnelIDVal.Store(false)
+	includeSiteLabelVal.Store(false)
+}

internal/telemetry/state_view.go

@@ -0,0 +1,106 @@
+package telemetry
+
+import (
+	"context"
+	"sync/atomic"
+	"time"
+
+	"go.opentelemetry.io/otel/attribute"
+	"go.opentelemetry.io/otel/metric"
+)
+
+// StateView provides a read-only view for observable gauges.
+// Implementations must be concurrency-safe and avoid blocking operations.
+// All methods should be fast and use RLocks where applicable.
+type StateView interface {
+	// ListSites returns a stable, low-cardinality list of site IDs to expose.
+	ListSites() []string
+	// Online returns whether the site is online.
+	Online(siteID string) (online bool, ok bool)
+	// LastHeartbeat returns the last heartbeat time for a site.
+	LastHeartbeat(siteID string) (t time.Time, ok bool)
+	// ActiveSessions returns the current number of active sessions for a site (across tunnels),
+	// or scoped to site if your model is site-scoped.
+	ActiveSessions(siteID string) (n int64, ok bool)
+}
+
+var (
+	stateView atomic.Value // of type StateView
+)
+
+// RegisterStateView sets the global StateView used by the default observable callback.
+func RegisterStateView(v StateView) {
+	stateView.Store(v)
+	// If instruments are registered, ensure a callback exists.
+	if v != nil {
+		SetObservableCallback(func(ctx context.Context, o metric.Observer) error {
+			if any := stateView.Load(); any != nil {
+				if sv, ok := any.(StateView); ok {
+					for _, siteID := range sv.ListSites() {
+						observeSiteOnlineFor(o, sv, siteID)
+						observeLastHeartbeatFor(o, sv, siteID)
+						observeSessionsFor(o, siteID, sv)
+					}
+				}
+			}
+			return nil
+		})
+	}
+}
+
+func observeSiteOnlineFor(o metric.Observer, sv StateView, siteID string) {
+	if online, ok := sv.Online(siteID); ok {
+		val := int64(0)
+		if online {
+			val = 1
+		}
+		o.ObserveInt64(mSiteOnline, val, metric.WithAttributes(
+			attribute.String("site_id", siteID),
+		))
+	}
+}
+
+func observeLastHeartbeatFor(o metric.Observer, sv StateView, siteID string) {
+	if t, ok := sv.LastHeartbeat(siteID); ok {
+		ts := float64(t.UnixNano()) / 1e9
+		o.ObserveFloat64(mSiteLastHeartbeat, ts, metric.WithAttributes(
+			attribute.String("site_id", siteID),
+		))
+	}
+}
+
+func observeSessionsFor(o metric.Observer, siteID string, any interface{}) {
+	if tm, ok := any.(interface{ SessionsByTunnel() map[string]int64 }); ok {
+		sessions := tm.SessionsByTunnel()
+		// If tunnel_id labels are enabled, preserve existing per-tunnel observations
+		if ShouldIncludeTunnelID() {
+			for tid, n := range sessions {
+				attrs := []attribute.KeyValue{
+					attribute.String("site_id", siteID),
+				}
+				if tid != "" {
+					attrs = append(attrs, attribute.String("tunnel_id", tid))
+				}
+				o.ObserveInt64(mTunnelSessions, n, metric.WithAttributes(attrs...))
+			}
+			return
+		}
+		// When tunnel_id is disabled, collapse per-tunnel counts into a single site-level value
+		var total int64
+		for _, n := range sessions {
+			total += n
+		}
+		// If there are no per-tunnel entries, fall back to ActiveSessions() if available
+		if total == 0 {
+			if svAny := stateView.Load(); svAny != nil {
+				if sv, ok := svAny.(StateView); ok {
+					if n, ok2 := sv.ActiveSessions(siteID); ok2 {
+						total = n
+					}
+				}
+			}
+		}
+		o.ObserveInt64(mTunnelSessions, total, metric.WithAttributes(attribute.String("site_id", siteID)))
+		return
+	}
+}

internal/telemetry/telemetry.go

@@ -0,0 +1,384 @@
+package telemetry
+
+import (
+	"context"
+	"errors"
+	"net/http"
+	"os"
+	"strings"
+	"sync/atomic"
+	"time"
+
+	promclient "github.com/prometheus/client_golang/prometheus"
+	"github.com/prometheus/client_golang/prometheus/promhttp"
+	"go.opentelemetry.io/contrib/instrumentation/runtime"
+	"go.opentelemetry.io/otel"
+	"go.opentelemetry.io/otel/attribute"
+	"go.opentelemetry.io/otel/exporters/otlp/otlpmetric/otlpmetricgrpc"
+	"go.opentelemetry.io/otel/exporters/otlp/otlptrace/otlptracegrpc"
+	"go.opentelemetry.io/otel/exporters/prometheus"
+	"go.opentelemetry.io/otel/sdk/metric"
+	"go.opentelemetry.io/otel/sdk/resource"
+	"go.opentelemetry.io/otel/sdk/trace"
+	semconv "go.opentelemetry.io/otel/semconv/v1.26.0"
+	"google.golang.org/grpc/credentials"
+)
+
+// Config controls telemetry initialization via env flags.
+//
+// Defaults align with the issue requirements:
+//   - Prometheus exporter enabled by default (/metrics)
+//   - OTLP exporter disabled by default
+//   - Durations in seconds, bytes in raw bytes
+//   - Admin HTTP server address configurable (for mounting /metrics)
+type Config struct {
+	ServiceName    string
+	ServiceVersion string
+
+	// Optional resource attributes
+	SiteID string
+	Region string
+
+	PromEnabled bool
+	OTLPEnabled bool
+
+	OTLPEndpoint string // host:port
+	OTLPInsecure bool
+
+	MetricExportInterval time.Duration
+	AdminAddr            string // e.g.: ":2112"
+
+	// Optional build info for newt_build_info metric
+	BuildVersion string
+	BuildCommit  string
+}
+
+// FromEnv reads configuration from environment variables.
+//
+//	NEWT_METRICS_PROMETHEUS_ENABLED (default: true)
+//	NEWT_METRICS_OTLP_ENABLED       (default: false)
+//	OTEL_EXPORTER_OTLP_ENDPOINT     (default: "localhost:4317")
+//	OTEL_EXPORTER_OTLP_INSECURE     (default: true)
+//	OTEL_METRIC_EXPORT_INTERVAL     (default: 15s)
+//	OTEL_SERVICE_NAME               (default: "newt")
+//	OTEL_SERVICE_VERSION            (default: "")
+//	NEWT_ADMIN_ADDR                 (default: ":2112")
+func FromEnv() Config {
+	// Prefer explicit NEWT_* env vars, then fall back to OTEL_RESOURCE_ATTRIBUTES
+	site := getenv("NEWT_SITE_ID", "")
+	if site == "" {
+		site = getenv("NEWT_ID", "")
+	}
+	region := os.Getenv("NEWT_REGION")
+	if site == "" || region == "" {
+		if ra := os.Getenv("OTEL_RESOURCE_ATTRIBUTES"); ra != "" {
+			m := parseResourceAttributes(ra)
+			if site == "" {
+				site = m["site_id"]
+			}
+			if region == "" {
+				region = m["region"]
+			}
+		}
+	}
+	return Config{
+		ServiceName:          getenv("OTEL_SERVICE_NAME", "newt"),
+		ServiceVersion:       os.Getenv("OTEL_SERVICE_VERSION"),
+		SiteID:               site,
+		Region:               region,
+		PromEnabled:          getenv("NEWT_METRICS_PROMETHEUS_ENABLED", "true") == "true",
+		OTLPEnabled:          getenv("NEWT_METRICS_OTLP_ENABLED", "false") == "true",
+		OTLPEndpoint:         getenv("OTEL_EXPORTER_OTLP_ENDPOINT", "localhost:4317"),
+		OTLPInsecure:         getenv("OTEL_EXPORTER_OTLP_INSECURE", "true") == "true",
+		MetricExportInterval: getdur("OTEL_METRIC_EXPORT_INTERVAL", 15*time.Second),
+		AdminAddr:            getenv("NEWT_ADMIN_ADDR", ":2112"),
+	}
+}
+
+// Setup holds initialized telemetry providers and (optionally) a /metrics handler.
+// Call Shutdown when the process terminates to flush exporters.
+type Setup struct {
+	MeterProvider  *metric.MeterProvider
+	TracerProvider *trace.TracerProvider
+
+	PrometheusHandler http.Handler // nil if Prometheus exporter disabled
+
+	shutdowns []func(context.Context) error
+}
+
+// Init configures OpenTelemetry metrics and (optionally) tracing.
+//
+// It sets a global MeterProvider and TracerProvider, registers runtime instrumentation,
+// installs recommended histogram views for *_latency_seconds, and returns a Setup with
+// a Shutdown method to flush exporters.
+func Init(ctx context.Context, cfg Config) (*Setup, error) {
+	// Configure tunnel_id label inclusion from env (default true)
+	if getenv("NEWT_METRICS_INCLUDE_TUNNEL_ID", "true") == "true" {
+		includeTunnelIDVal.Store(true)
+	} else {
+		includeTunnelIDVal.Store(false)
+	}
+        if getenv("NEWT_METRICS_INCLUDE_SITE_LABELS", "true") == "true" {
+                includeSiteLabelVal.Store(true)
+        } else {
+                includeSiteLabelVal.Store(false)
+        }
+	res := buildResource(ctx, cfg)
+	UpdateSiteInfo(cfg.SiteID, cfg.Region)
+
+	s := &Setup{}
+	readers, promHandler, shutdowns, err := setupMetricExport(ctx, cfg, res)
+	if err != nil {
+		return nil, err
+	}
+	s.PrometheusHandler = promHandler
+	// Build provider
+	mp := buildMeterProvider(res, readers)
+	otel.SetMeterProvider(mp)
+	s.MeterProvider = mp
+	s.shutdowns = append(s.shutdowns, mp.Shutdown)
+	// Optional tracing
+	if cfg.OTLPEnabled {
+		if tp, shutdown := setupTracing(ctx, cfg, res); tp != nil {
+			otel.SetTracerProvider(tp)
+			s.TracerProvider = tp
+			s.shutdowns = append(s.shutdowns, func(c context.Context) error {
+				return errors.Join(shutdown(c), tp.Shutdown(c))
+			})
+		}
+	}
+	// Add metric exporter shutdowns
+	s.shutdowns = append(s.shutdowns, shutdowns...)
+	// Runtime metrics
+	_ = runtime.Start(runtime.WithMeterProvider(mp))
+	// Instruments
+	if err := registerInstruments(); err != nil {
+		return nil, err
+	}
+	if cfg.BuildVersion != "" || cfg.BuildCommit != "" {
+		RegisterBuildInfo(cfg.BuildVersion, cfg.BuildCommit)
+	}
+	return s, nil
+}
+
+func buildResource(ctx context.Context, cfg Config) *resource.Resource {
+	attrs := []attribute.KeyValue{
+		semconv.ServiceName(cfg.ServiceName),
+		semconv.ServiceVersion(cfg.ServiceVersion),
+	}
+	if cfg.SiteID != "" {
+		attrs = append(attrs, attribute.String("site_id", cfg.SiteID))
+	}
+	if cfg.Region != "" {
+		attrs = append(attrs, attribute.String("region", cfg.Region))
+	}
+	res, _ := resource.New(ctx, resource.WithFromEnv(), resource.WithHost(), resource.WithAttributes(attrs...))
+	return res
+}
+
+func setupMetricExport(ctx context.Context, cfg Config, _ *resource.Resource) ([]metric.Reader, http.Handler, []func(context.Context) error, error) {
+	var readers []metric.Reader
+	var shutdowns []func(context.Context) error
+	var promHandler http.Handler
+	if cfg.PromEnabled {
+		reg := promclient.NewRegistry()
+		exp, err := prometheus.New(prometheus.WithRegisterer(reg))
+		if err != nil {
+			return nil, nil, nil, err
+		}
+		readers = append(readers, exp)
+		promHandler = promhttp.HandlerFor(reg, promhttp.HandlerOpts{})
+	}
+	if cfg.OTLPEnabled {
+		mopts := []otlpmetricgrpc.Option{otlpmetricgrpc.WithEndpoint(cfg.OTLPEndpoint)}
+		if hdrs := parseOTLPHeaders(os.Getenv("OTEL_EXPORTER_OTLP_HEADERS")); len(hdrs) > 0 {
+			mopts = append(mopts, otlpmetricgrpc.WithHeaders(hdrs))
+		}
+		if cfg.OTLPInsecure {
+			mopts = append(mopts, otlpmetricgrpc.WithInsecure())
+		} else if certFile := os.Getenv("OTEL_EXPORTER_OTLP_CERTIFICATE"); certFile != "" {
+			if creds, cerr := credentials.NewClientTLSFromFile(certFile, ""); cerr == nil {
+				mopts = append(mopts, otlpmetricgrpc.WithTLSCredentials(creds))
+			}
+		}
+		mexp, err := otlpmetricgrpc.New(ctx, mopts...)
+		if err != nil {
+			return nil, nil, nil, err
+		}
+		readers = append(readers, metric.NewPeriodicReader(mexp, metric.WithInterval(cfg.MetricExportInterval)))
+		shutdowns = append(shutdowns, mexp.Shutdown)
+	}
+	return readers, promHandler, shutdowns, nil
+}
+
+func buildMeterProvider(res *resource.Resource, readers []metric.Reader) *metric.MeterProvider {
+	var mpOpts []metric.Option
+	mpOpts = append(mpOpts, metric.WithResource(res))
+	for _, r := range readers {
+		mpOpts = append(mpOpts, metric.WithReader(r))
+	}
+	mpOpts = append(mpOpts, metric.WithView(metric.NewView(
+		metric.Instrument{Name: "newt_*_latency_seconds"},
+		metric.Stream{Aggregation: metric.AggregationExplicitBucketHistogram{Boundaries: []float64{0.005, 0.01, 0.025, 0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 10, 30}}},
+	)))
+	mpOpts = append(mpOpts, metric.WithView(metric.NewView(
+		metric.Instrument{Name: "newt_*"},
+		metric.Stream{AttributeFilter: func(kv attribute.KeyValue) bool {
+			k := string(kv.Key)
+			switch k {
+			case "tunnel_id", "transport", "direction", "protocol", "result", "reason", "initiator", "error_type", "msg_type", "phase", "version", "commit", "site_id", "region":
+				return true
+			default:
+				return false
+			}
+		}},
+	)))
+	return metric.NewMeterProvider(mpOpts...)
+}
+
+func setupTracing(ctx context.Context, cfg Config, res *resource.Resource) (*trace.TracerProvider, func(context.Context) error) {
+	topts := []otlptracegrpc.Option{otlptracegrpc.WithEndpoint(cfg.OTLPEndpoint)}
+	if hdrs := parseOTLPHeaders(os.Getenv("OTEL_EXPORTER_OTLP_HEADERS")); len(hdrs) > 0 {
+		topts = append(topts, otlptracegrpc.WithHeaders(hdrs))
+	}
+	if cfg.OTLPInsecure {
+		topts = append(topts, otlptracegrpc.WithInsecure())
+	} else if certFile := os.Getenv("OTEL_EXPORTER_OTLP_CERTIFICATE"); certFile != "" {
+		if creds, cerr := credentials.NewClientTLSFromFile(certFile, ""); cerr == nil {
+			topts = append(topts, otlptracegrpc.WithTLSCredentials(creds))
+		}
+	}
+	exp, err := otlptracegrpc.New(ctx, topts...)
+	if err != nil {
+		return nil, nil
+	}
+	tp := trace.NewTracerProvider(trace.WithBatcher(exp), trace.WithResource(res))
+	return tp, exp.Shutdown
+}
+
+// Shutdown flushes exporters and providers in reverse init order.
+func (s *Setup) Shutdown(ctx context.Context) error {
+	var err error
+	for i := len(s.shutdowns) - 1; i >= 0; i-- {
+		err = errors.Join(err, s.shutdowns[i](ctx))
+	}
+	return err
+}
+
+func parseOTLPHeaders(h string) map[string]string {
+	m := map[string]string{}
+	if h == "" {
+		return m
+	}
+	pairs := strings.Split(h, ",")
+	for _, p := range pairs {
+		kv := strings.SplitN(strings.TrimSpace(p), "=", 2)
+		if len(kv) == 2 {
+			m[strings.TrimSpace(kv[0])] = strings.TrimSpace(kv[1])
+		}
+	}
+	return m
+}
+
+// parseResourceAttributes parses OTEL_RESOURCE_ATTRIBUTES formatted as k=v,k2=v2
+func parseResourceAttributes(s string) map[string]string {
+	m := map[string]string{}
+	if s == "" {
+		return m
+	}
+	parts := strings.Split(s, ",")
+	for _, p := range parts {
+		kv := strings.SplitN(strings.TrimSpace(p), "=", 2)
+		if len(kv) == 2 {
+			m[strings.TrimSpace(kv[0])] = strings.TrimSpace(kv[1])
+		}
+	}
+	return m
+}
+
+// Global site/region used to enrich metric labels.
+var siteIDVal atomic.Value
+var regionVal atomic.Value
+var (
+	includeTunnelIDVal  atomic.Value // bool; default true
+	includeSiteLabelVal atomic.Value // bool; default false
+)
+
+// UpdateSiteInfo updates the global site_id and region used for metric labels.
+// Thread-safe via atomic.Value: subsequent metric emissions will include
+// the new labels, prior emissions remain unchanged.
+func UpdateSiteInfo(siteID, region string) {
+	if siteID != "" {
+		siteIDVal.Store(siteID)
+	}
+	if region != "" {
+		regionVal.Store(region)
+	}
+}
+
+func getSiteID() string {
+	if v, ok := siteIDVal.Load().(string); ok {
+		return v
+	}
+	return ""
+}
+
+func getRegion() string {
+	if v, ok := regionVal.Load().(string); ok {
+		return v
+	}
+	return ""
+}
+
+// siteAttrs returns label KVs for site_id and region (if set).
+func siteAttrs() []attribute.KeyValue {
+	var out []attribute.KeyValue
+	if s := getSiteID(); s != "" {
+		out = append(out, attribute.String("site_id", s))
+	}
+	if r := getRegion(); r != "" {
+		out = append(out, attribute.String("region", r))
+	}
+	return out
+}
+
+// SiteLabelKVs exposes site label KVs for other packages (e.g., proxy manager).
+func SiteLabelKVs() []attribute.KeyValue {
+	if !ShouldIncludeSiteLabels() {
+		return nil
+	}
+	return siteAttrs()
+}
+
+// ShouldIncludeTunnelID returns whether tunnel_id labels should be emitted.
+func ShouldIncludeTunnelID() bool {
+	if v, ok := includeTunnelIDVal.Load().(bool); ok {
+		return v
+	}
+	return true
+}
+
+// ShouldIncludeSiteLabels returns whether site_id/region should be emitted as
+// metric labels in addition to resource attributes.
+func ShouldIncludeSiteLabels() bool {
+	if v, ok := includeSiteLabelVal.Load().(bool); ok {
+		return v
+	}
+	return false
+}
+
+func getenv(k, d string) string {
+	if v := os.Getenv(k); v != "" {
+		return v
+	}
+	return d
+}
+
+func getdur(k string, d time.Duration) time.Duration {
+	if v := os.Getenv(k); v != "" {
+		if p, e := time.ParseDuration(v); e == nil {
+			return p
+		}
+	}
+	return d
+}

internal/telemetry/telemetry_attrfilter_test.go

@@ -0,0 +1,53 @@
+package telemetry
+
+import (
+	"context"
+	"io"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"testing"
+	"time"
+
+	"go.opentelemetry.io/otel/attribute"
+)
+
+// Test that disallowed attributes are filtered from the exposition.
+func TestAttributeFilterDropsUnknownKeys(t *testing.T) {
+        ctx := context.Background()
+        resetMetricsForTest()
+        t.Setenv("NEWT_METRICS_INCLUDE_SITE_LABELS", "true")
+        cfg := Config{ServiceName: "newt", PromEnabled: true, AdminAddr: "127.0.0.1:0"}
+	tel, err := Init(ctx, cfg)
+	if err != nil {
+		t.Fatalf("init: %v", err)
+	}
+	defer func() { _ = tel.Shutdown(context.Background()) }()
+
+	if tel.PrometheusHandler == nil {
+		t.Fatalf("prom handler nil")
+	}
+	ts := httptest.NewServer(tel.PrometheusHandler)
+	defer ts.Close()
+
+	// Add samples with disallowed attribute keys
+	for _, k := range []string{"forbidden", "site_id", "host"} {
+		set := attribute.NewSet(attribute.String(k, "x"))
+		AddTunnelBytesSet(ctx, 123, set)
+	}
+	time.Sleep(50 * time.Millisecond)
+
+	resp, err := http.Get(ts.URL)
+	if err != nil {
+		t.Fatalf("GET: %v", err)
+	}
+	defer resp.Body.Close()
+	b, _ := io.ReadAll(resp.Body)
+	body := string(b)
+	if strings.Contains(body, "forbidden=") {
+		t.Fatalf("unexpected forbidden attribute leaked into metrics: %s", body)
+	}
+	if !strings.Contains(body, "site_id=\"x\"") {
+		t.Fatalf("expected allowed attribute site_id to be present in metrics, got: %s", body)
+	}
+}

internal/telemetry/telemetry_golden_test.go

@@ -0,0 +1,76 @@
+package telemetry
+
+import (
+	"bufio"
+	"context"
+	"io"
+	"net/http"
+	"net/http/httptest"
+	"os"
+	"path/filepath"
+	"strings"
+	"testing"
+	"time"
+)
+
+// Golden test that /metrics contains expected metric names.
+func TestMetricsGoldenContains(t *testing.T) {
+	ctx := context.Background()
+	resetMetricsForTest()
+	t.Setenv("NEWT_METRICS_INCLUDE_SITE_LABELS", "true")
+	cfg := Config{ServiceName: "newt", PromEnabled: true, AdminAddr: "127.0.0.1:0", BuildVersion: "test"}
+	tel, err := Init(ctx, cfg)
+	if err != nil {
+		t.Fatalf("telemetry init error: %v", err)
+	}
+	defer func() { _ = tel.Shutdown(context.Background()) }()
+
+	if tel.PrometheusHandler == nil {
+		t.Fatalf("prom handler nil")
+	}
+	ts := httptest.NewServer(tel.PrometheusHandler)
+	defer ts.Close()
+
+	// Trigger counters to ensure they appear in the scrape
+	IncConnAttempt(ctx, "websocket", "success")
+	IncWSReconnect(ctx, "io_error")
+	IncProxyConnectionEvent(ctx, "", "tcp", ProxyConnectionOpened)
+	if tel.MeterProvider != nil {
+		_ = tel.MeterProvider.ForceFlush(ctx)
+	}
+	time.Sleep(100 * time.Millisecond)
+
+	var body string
+	for i := 0; i < 5; i++ {
+		resp, err := http.Get(ts.URL)
+		if err != nil {
+			t.Fatalf("GET metrics failed: %v", err)
+		}
+		b, _ := io.ReadAll(resp.Body)
+		_ = resp.Body.Close()
+		body = string(b)
+		if strings.Contains(body, "newt_connection_attempts_total") {
+			break
+		}
+		time.Sleep(100 * time.Millisecond)
+	}
+
+	f, err := os.Open(filepath.Join("testdata", "expected_contains.golden"))
+	if err != nil {
+		t.Fatalf("read golden: %v", err)
+	}
+	defer f.Close()
+	s := bufio.NewScanner(f)
+	for s.Scan() {
+		needle := strings.TrimSpace(s.Text())
+		if needle == "" {
+			continue
+		}
+		if !strings.Contains(body, needle) {
+			t.Fatalf("expected metrics body to contain %q. body=\n%s", needle, body)
+		}
+	}
+	if err := s.Err(); err != nil {
+		t.Fatalf("scan golden: %v", err)
+	}
+}

internal/telemetry/telemetry_smoke_test.go

@@ -0,0 +1,65 @@
+package telemetry
+
+import (
+	"context"
+	"io"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"testing"
+	"time"
+)
+
+// Smoke test that /metrics contains at least one newt_* metric when Prom exporter is enabled.
+func TestMetricsSmoke(t *testing.T) {
+	ctx := context.Background()
+	resetMetricsForTest()
+	t.Setenv("NEWT_METRICS_INCLUDE_SITE_LABELS", "true")
+	cfg := Config{
+		ServiceName:          "newt",
+		PromEnabled:          true,
+		OTLPEnabled:          false,
+		AdminAddr:            "127.0.0.1:0",
+		BuildVersion:         "test",
+		BuildCommit:          "deadbeef",
+		MetricExportInterval: 5 * time.Second,
+	}
+	tel, err := Init(ctx, cfg)
+	if err != nil {
+		t.Fatalf("telemetry init error: %v", err)
+	}
+	defer func() { _ = tel.Shutdown(context.Background()) }()
+
+	// Serve the Prom handler on a test server
+	if tel.PrometheusHandler == nil {
+		t.Fatalf("Prometheus handler nil; PromEnabled should enable it")
+	}
+	ts := httptest.NewServer(tel.PrometheusHandler)
+	defer ts.Close()
+
+	// Record a simple metric and then fetch /metrics
+	IncConnAttempt(ctx, "websocket", "success")
+	if tel.MeterProvider != nil {
+		_ = tel.MeterProvider.ForceFlush(ctx)
+	}
+	// Give the exporter a tick to collect
+	time.Sleep(100 * time.Millisecond)
+
+	var body string
+	for i := 0; i < 5; i++ {
+		resp, err := http.Get(ts.URL)
+		if err != nil {
+			t.Fatalf("GET /metrics failed: %v", err)
+		}
+		b, _ := io.ReadAll(resp.Body)
+		_ = resp.Body.Close()
+		body = string(b)
+		if strings.Contains(body, "newt_connection_attempts_total") {
+			break
+		}
+		time.Sleep(100 * time.Millisecond)
+	}
+	if !strings.Contains(body, "newt_connection_attempts_total") {
+		t.Fatalf("expected newt_connection_attempts_total in metrics, got:\n%s", body)
+	}
+}

internal/telemetry/testdata/expected_contains.golden

@@ -0,0 +1,3 @@
+newt_connection_attempts_total
+newt_websocket_reconnects_total
+newt_proxy_connections_total

logger/logger.go

@@ -2,15 +2,15 @@ package logger
 
 import (
 	"fmt"
-	"log"
 	"os"
+	"strings"
 	"sync"
 	"time"
 )
 
 // Logger struct holds the logger instance
 type Logger struct {
-	logger *log.Logger
+	writer LogWriter
 	level  LogLevel
 }
 
@@ -19,17 +19,29 @@ var (
 	once          sync.Once
 )
 
-// NewLogger creates a new logger instance
+// NewLogger creates a new logger instance with the default StandardWriter
 func NewLogger() *Logger {
 	return &Logger{
-		logger: log.New(os.Stdout, "", 0),
+		writer: NewStandardWriter(),
+		level:  DEBUG,
+	}
+}
+
+// NewLoggerWithWriter creates a new logger instance with a custom LogWriter
+func NewLoggerWithWriter(writer LogWriter) *Logger {
+	return &Logger{
+		writer: writer,
 		level:  DEBUG,
 	}
 }
 
 // Init initializes the default logger
-func Init() *Logger {
+func Init(logger *Logger) *Logger {
 	once.Do(func() {
+		if logger != nil {
+			defaultLogger = logger
+			return
+		}
 		defaultLogger = NewLogger()
 	})
 	return defaultLogger
@@ -38,7 +50,7 @@ func Init() *Logger {
 // GetLogger returns the default logger instance
 func GetLogger() *Logger {
 	if defaultLogger == nil {
-		Init()
+		Init(nil)
 	}
 	return defaultLogger
 }
@@ -48,14 +60,21 @@ func (l *Logger) SetLevel(level LogLevel) {
 	l.level = level
 }
 
+// SetOutput sets the output destination for the logger (only works with StandardWriter)
+func (l *Logger) SetOutput(output *os.File) {
+	if sw, ok := l.writer.(*StandardWriter); ok {
+		sw.SetOutput(output)
+	}
+}
+
 // log handles the actual logging
 func (l *Logger) log(level LogLevel, format string, args ...interface{}) {
 	if level < l.level {
 		return
 	}
-	timestamp := time.Now().Format("2006/01/02 15:04:05")
+
 	message := fmt.Sprintf(format, args...)
-	l.logger.Printf("%s: %s %s", level.String(), timestamp, message)
+	l.writer.Write(level, time.Now(), message)
 }
 
 // Debug logs debug level messages
@@ -104,3 +123,31 @@ func Error(format string, args ...interface{}) {
 func Fatal(format string, args ...interface{}) {
 	GetLogger().Fatal(format, args...)
 }
+
+// SetOutput sets the output destination for the default logger
+func SetOutput(output *os.File) {
+	GetLogger().SetOutput(output)
+}
+
+// WireGuardLogger is a wrapper type that matches WireGuard's Logger interface
+type WireGuardLogger struct {
+	Verbosef func(format string, args ...any)
+	Errorf   func(format string, args ...any)
+}
+
+// GetWireGuardLogger returns a WireGuard-compatible logger that writes to the newt logger
+// The prepend string is added as a prefix to all log messages
+func (l *Logger) GetWireGuardLogger(prepend string) *WireGuardLogger {
+	return &WireGuardLogger{
+		Verbosef: func(format string, args ...any) {
+			// if the format string contains "Sending keepalive packet", skip debug logging to reduce noise
+			if strings.Contains(format, "Sending keepalive packet") {
+				return
+			}
+			l.Debug(prepend+format, args...)
+		},
+		Errorf: func(format string, args ...any) {
+			l.Error(prepend+format, args...)
+		},
+	}
+}

logger/writer.go

@@ -0,0 +1,54 @@
+package logger
+
+import (
+	"fmt"
+	"os"
+	"time"
+)
+
+// LogWriter is an interface for writing log messages
+// Implement this interface to create custom log backends (OS log, syslog, etc.)
+type LogWriter interface {
+	// Write writes a log message with the given level, timestamp, and formatted message
+	Write(level LogLevel, timestamp time.Time, message string)
+}
+
+// StandardWriter is the default log writer that writes to an io.Writer
+type StandardWriter struct {
+	output   *os.File
+	timezone *time.Location
+}
+
+// NewStandardWriter creates a new standard writer with the default configuration
+func NewStandardWriter() *StandardWriter {
+	// Get timezone from environment variable or use local timezone
+	timezone := os.Getenv("LOGGER_TIMEZONE")
+	var location *time.Location
+	var err error
+
+	if timezone != "" {
+		location, err = time.LoadLocation(timezone)
+		if err != nil {
+			// If invalid timezone, fall back to local
+			location = time.Local
+		}
+	} else {
+		location = time.Local
+	}
+
+	return &StandardWriter{
+		output:   os.Stdout,
+		timezone: location,
+	}
+}
+
+// SetOutput sets the output destination
+func (w *StandardWriter) SetOutput(output *os.File) {
+	w.output = output
+}
+
+// Write implements the LogWriter interface
+func (w *StandardWriter) Write(level LogLevel, timestamp time.Time, message string) {
+	formattedTime := timestamp.In(w.timezone).Format("2006/01/02 15:04:05")
+	fmt.Fprintf(w.output, "%s: %s %s\n", level.String(), formattedTime, message)
+}

nativessh/auth.go

@@ -0,0 +1,158 @@
+package nativessh
+
+import (
+	"bufio"
+	"fmt"
+	"net"
+	"os"
+	"os/user"
+	"path/filepath"
+	"strings"
+
+	"github.com/fosrl/newt/logger"
+	"golang.org/x/crypto/ssh"
+)
+
+type staticConnMeta struct {
+	user string
+}
+
+func (m staticConnMeta) User() string          { return m.user }
+func (m staticConnMeta) SessionID() []byte     { return nil }
+func (m staticConnMeta) ClientVersion() []byte { return nil }
+func (m staticConnMeta) ServerVersion() []byte { return nil }
+func (m staticConnMeta) RemoteAddr() net.Addr  { return nil }
+func (m staticConnMeta) LocalAddr() net.Addr   { return nil }
+
+// CheckAuthorizedKeys reports whether key matches any entry in the system
+// user's ~/.ssh/authorized_keys file.  Returns false (not an error) when the
+// user or file does not exist.
+func CheckAuthorizedKeys(username string, key ssh.PublicKey) bool {
+	u, err := user.Lookup(username)
+	if err != nil {
+		return false
+	}
+	f, err := os.Open(filepath.Join(u.HomeDir, ".ssh", "authorized_keys"))
+	if err != nil {
+		return false
+	}
+	defer f.Close()
+
+	want := ssh.FingerprintSHA256(key)
+	scanner := bufio.NewScanner(f)
+	for scanner.Scan() {
+		line := strings.TrimSpace(scanner.Text())
+		if line == "" || strings.HasPrefix(line, "#") {
+			continue
+		}
+		parsed, _, _, _, err := ssh.ParseAuthorizedKey([]byte(line))
+		if err != nil {
+			continue
+		}
+		if ssh.FingerprintSHA256(parsed) == want {
+			return true
+		}
+	}
+	return false
+}
+
+// SystemUserExists reports whether a user account with the given name exists
+// on the host OS.
+func SystemUserExists(username string) bool {
+	_, err := user.Lookup(username)
+	return err == nil
+}
+
+// Authenticate authenticates a user for a browser-based native SSH session.
+// It tries, in order:
+//  1. Private key — parses privateKeyPEM and checks it against the user's
+//     ~/.ssh/authorized_keys.
+//  2. Password — verifies password via the host OS PAM stack (Linux only).
+//
+// Returns nil on the first method that succeeds, or an error if all fail.
+func Authenticate(username, password, privateKeyPEM string) error {
+	return AuthenticateWithCertificate(nil, username, password, privateKeyPEM, "")
+}
+
+// AuthenticateWithCertificate authenticates a user for a browser-based native
+// SSH session using the same method ordering as the native SSH server:
+//  1. Private key in host ~/.ssh/authorized_keys.
+//  2. SSH certificate signed by the configured CA (when provided).
+//  3. Password via host PAM stack.
+func AuthenticateWithCertificate(store *CredentialStore, username, password, privateKeyPEM, certificate string) error {
+	logger.Debug("nativessh: authenticating user %q (hasPassword=%v, hasPrivateKey=%v)", username, password != "", privateKeyPEM != "")
+	if !SystemUserExists(username) {
+		logger.Debug("nativessh: user %q not found on system", username)
+		return fmt.Errorf("user %q does not exist", username)
+	}
+
+	var signer ssh.Signer
+	if privateKeyPEM != "" {
+		parsedSigner, err := ssh.ParsePrivateKey([]byte(privateKeyPEM))
+		if err != nil {
+			logger.Debug("nativessh: failed to parse private key for %q: %v", username, err)
+		} else if CheckAuthorizedKeys(username, parsedSigner.PublicKey()) {
+			logger.Debug("nativessh: private key auth succeeded for %q", username)
+			return nil
+		} else {
+			signer = parsedSigner
+			logger.Debug("nativessh: private key not in authorized_keys for %q", username)
+		}
+	}
+
+	if store != nil && certificate != "" {
+		if signer == nil {
+			logger.Debug("nativessh: certificate provided for %q but no matching private key was provided", username)
+		} else {
+			pub, _, _, _, err := ssh.ParseAuthorizedKey([]byte(certificate))
+			if err != nil {
+				logger.Debug("nativessh: failed to parse certificate for %q: %v", username, err)
+			} else {
+				cert, ok := pub.(*ssh.Certificate)
+				if !ok {
+					logger.Debug("nativessh: provided cert data for %q is not an SSH certificate", username)
+				} else if ssh.FingerprintSHA256(cert.Key) != ssh.FingerprintSHA256(signer.PublicKey()) {
+					logger.Debug("nativessh: certificate key mismatch for %q", username)
+				} else {
+					caKey, userPrincipals := store.get(username)
+					if caKey == nil {
+						logger.Debug("nativessh: CA key is not set for certificate auth user %q", username)
+					} else if len(userPrincipals) == 0 {
+						logger.Debug("nativessh: no allowed principals found for certificate auth user %q", username)
+					} else {
+						checker := &ssh.CertChecker{
+							IsUserAuthority: func(auth ssh.PublicKey) bool {
+								return ssh.FingerprintSHA256(auth) == ssh.FingerprintSHA256(caKey)
+							},
+						}
+
+						var lastErr error
+						for principal := range userPrincipals {
+							_, authErr := checker.Authenticate(staticConnMeta{user: principal}, cert)
+							if authErr == nil {
+								logger.Debug("nativessh: certificate auth succeeded for %q (principal=%q)", username, principal)
+								return nil
+							}
+							lastErr = authErr
+						}
+						if lastErr != nil {
+							logger.Debug("nativessh: certificate auth failed for %q: %v", username, lastErr)
+						}
+					}
+				}
+			}
+		}
+	}
+
+	if password != "" {
+		if err := VerifySystemPassword(username, password); err != nil {
+			logger.Debug("nativessh: password auth failed for %q: %v", username, err)
+		} else {
+			logger.Debug("nativessh: password auth succeeded for %q", username)
+			return nil
+		}
+	} else {
+		logger.Debug("nativessh: no password provided for %q", username)
+	}
+	return fmt.Errorf("authentication failed for user %q", username)
+}

nativessh/pam_linux.go

@@ -0,0 +1,99 @@
+//go:build linux
+
+package nativessh
+
+import (
+	"bufio"
+	"bytes"
+	"errors"
+	"fmt"
+	"os"
+	"strings"
+
+	"github.com/fosrl/newt/logger"
+	"github.com/go-crypt/crypt"
+	"github.com/go-crypt/x/yescrypt"
+)
+
+// VerifySystemPassword authenticates username/password by reading /etc/shadow.
+// Supports yescrypt ($y$), bcrypt ($2b$/$2a$/$2y$), SHA-512 ($6$), SHA-256
+// ($5$), argon2, scrypt, and other schemes handled by go-crypt/crypt.
+func VerifySystemPassword(username, password string) error {
+	hash, err := readShadowHash(username)
+	if err != nil {
+		logger.Debug("nativessh/pam: readShadowHash for %q failed: %v", username, err)
+		return fmt.Errorf("shadow: %w", err)
+	}
+
+	// Log the scheme prefix only (never the full hash).
+	scheme := "unknown"
+	for _, prefix := range []string{"$y$", "$2a$", "$2b$", "$2y$", "$6$", "$5$", "$1$"} {
+		if strings.HasPrefix(hash, prefix) {
+			scheme = prefix
+			break
+		}
+	}
+	logger.Debug("nativessh/pam: verifying password for %q using scheme %s", username, scheme)
+
+	// Yescrypt ($y$) is not in go-crypt/crypt's default decoder; handle it directly.
+	if strings.HasPrefix(hash, "$y$") {
+		computed, err := yescrypt.Hash([]byte(password), []byte(hash))
+		if err != nil {
+			logger.Debug("nativessh/pam: yescrypt.Hash for %q failed: %v", username, err)
+			return fmt.Errorf("yescrypt: %w", err)
+		}
+		if !bytes.Equal(computed, []byte(hash)) {
+			logger.Debug("nativessh/pam: yescrypt mismatch for %q", username)
+			return errors.New("authentication failed")
+		}
+		return nil
+	}
+
+	decoder, err := crypt.NewDefaultDecoder()
+	if err != nil {
+		return fmt.Errorf("crypt decoder: %w", err)
+	}
+
+	digest, err := decoder.Decode(hash)
+	if err != nil {
+		logger.Debug("nativessh/pam: failed to decode hash for %q: %v", username, err)
+		return fmt.Errorf("unsupported password hash scheme %q: %w", scheme, err)
+	}
+
+	match, err := digest.MatchAdvanced(password)
+	if err != nil {
+		logger.Debug("nativessh/pam: MatchAdvanced for %q failed: %v", username, err)
+		return err
+	}
+	if !match {
+		logger.Debug("nativessh/pam: password mismatch for %q", username)
+		return errors.New("authentication failed")
+	}
+	return nil
+}
+
+// readShadowHash reads /etc/shadow and returns the password hash for username.
+func readShadowHash(username string) (string, error) {
+	f, err := os.Open("/etc/shadow")
+	if err != nil {
+		return "", err
+	}
+	defer f.Close()
+
+	scanner := bufio.NewScanner(f)
+	for scanner.Scan() {
+		fields := strings.SplitN(scanner.Text(), ":", 3)
+		if len(fields) < 2 || fields[0] != username {
+			continue
+		}
+		h := fields[1]
+		if h == "" || h == "*" || strings.HasPrefix(h, "!") || h == "x" {
+			return "", errors.New("account locked or has no password")
+		}
+		return h, nil
+	}
+	if err := scanner.Err(); err != nil {
+		return "", err
+	}
+	return "", errors.New("user not found in shadow database")
+}

nativessh/pam_other.go

@@ -0,0 +1,11 @@
+//go:build !linux
+
+package nativessh
+
+import "errors"
+
+// VerifySystemPassword is not supported on non-Linux platforms; it always
+// returns an error so that password authentication is never accepted.
+func VerifySystemPassword(username, password string) error {
+	return errors.New("password authentication not supported on this platform")
+}

nativessh/pty.go

@@ -0,0 +1,130 @@
+//go:build !windows
+
+package nativessh
+
+import (
+	"bufio"
+	"errors"
+	"fmt"
+	"os"
+	"os/exec"
+	"os/user"
+	"strings"
+	"sync"
+
+	"github.com/creack/pty"
+)
+
+// PTYSession is a running shell process attached to a PTY.
+// It implements io.ReadWriteCloser so it can be bridged to any transport.
+type PTYSession struct {
+	ptmx     *os.File
+	cmd      *exec.Cmd
+	waitOnce sync.Once
+	exitCode int
+}
+
+// findShell returns the path to the best available interactive shell by
+// checking preferred shells in order, falling back to /bin/sh.
+func findShell() string {
+	preferred := []string{"zsh", "bash", "fish", "ksh", "sh"}
+	for _, name := range preferred {
+		if path, err := exec.LookPath(name); err == nil {
+			return path
+		}
+	}
+	return "/bin/sh"
+}
+
+// userShell returns the login shell configured for u in /etc/passwd.
+// If the field is empty or the binary does not exist, it falls back to
+// findShell so there is always a usable shell.
+func userShell(u *user.User) string {
+	if shell := passwdShell(u.Username); shell != "" {
+		if _, err := exec.LookPath(shell); err == nil {
+			return shell
+		}
+	}
+	return findShell()
+}
+
+// passwdShell reads /etc/passwd and returns the login shell for the named user.
+// Returns "" if the user is not found or the file cannot be read.
+func passwdShell(username string) string {
+	f, err := os.Open("/etc/passwd")
+	if err != nil {
+		return ""
+	}
+	defer f.Close()
+	scanner := bufio.NewScanner(f)
+	for scanner.Scan() {
+		line := scanner.Text()
+		if line == "" || line[0] == '#' {
+			continue
+		}
+		// Fields: username:password:uid:gid:gecos:home:shell
+		fields := strings.SplitN(line, ":", 7)
+		if len(fields) == 7 && fields[0] == username {
+			return fields[6]
+		}
+	}
+	_ = scanner.Err()
+	return ""
+}
+
+// NewPTYSession spawns the best available shell in a PTY.
+func NewPTYSession() (*PTYSession, error) {
+	shell := findShell()
+	cmd := exec.Command(shell)
+	cmd.Env = append(os.Environ(), "TERM=xterm-256color")
+	ptmx, err := pty.Start(cmd)
+	if err != nil {
+		return nil, fmt.Errorf("pty start: %w", err)
+	}
+	return &PTYSession{ptmx: ptmx, cmd: cmd}, nil
+}
+
+// Read reads output from the PTY.
+func (p *PTYSession) Read(b []byte) (int, error) {
+	return p.ptmx.Read(b)
+}
+
+// Write writes input to the PTY.
+func (p *PTYSession) Write(b []byte) (int, error) {
+	return p.ptmx.Write(b)
+}
+
+// Resize changes the PTY window size.
+func (p *PTYSession) Resize(cols, rows uint16) error {
+	return pty.Setsize(p.ptmx, &pty.Winsize{Cols: cols, Rows: rows})
+}
+
+// wait waits for the child process to exit exactly once and records its exit
+// code. Safe to call concurrently or multiple times.
+func (p *PTYSession) wait() {
+	p.waitOnce.Do(func() {
+		err := p.cmd.Wait()
+		if err != nil {
+			var exitErr *exec.ExitError
+			if errors.As(err, &exitErr) {
+				p.exitCode = exitErr.ExitCode()
+				return
+			}
+			p.exitCode = 1
+		}
+	})
+}
+
+// ExitCode waits for the shell process to exit and returns its exit code.
+// It is safe to call before or after Close.
+func (p *PTYSession) ExitCode() int {
+	p.wait()
+	return p.exitCode
+}
+
+// Close closes the PTY and waits for the child process to exit.
+func (p *PTYSession) Close() error {
+	err := p.ptmx.Close()
+	p.wait()
+	return err
+}

nativessh/pty_unix.go

@@ -0,0 +1,78 @@
+//go:build !windows
+
+package nativessh
+
+import (
+	"fmt"
+	"os"
+	"os/exec"
+	"os/user"
+	"strconv"
+	"syscall"
+
+	"github.com/creack/pty"
+)
+
+// NewPTYSessionAs spawns an interactive shell in a PTY running as the given
+// system user.  The calling process must have sufficient privileges (typically
+// root / CAP_SETUID) to switch to a different UID/GID.
+func NewPTYSessionAs(username string) (*PTYSession, error) {
+	u, err := user.Lookup(username)
+	if err != nil {
+		return nil, fmt.Errorf("user lookup %q: %w", username, err)
+	}
+	uid, err := strconv.ParseUint(u.Uid, 10, 32)
+	if err != nil {
+		return nil, fmt.Errorf("parse uid: %w", err)
+	}
+	gid, err := strconv.ParseUint(u.Gid, 10, 32)
+	if err != nil {
+		return nil, fmt.Errorf("parse gid: %w", err)
+	}
+
+	// Collect supplementary group IDs.
+	groupIDs, err := u.GroupIds()
+	if err != nil {
+		groupIDs = []string{}
+	}
+	var groups []uint32
+	for _, g := range groupIDs {
+		gval, err := strconv.ParseUint(g, 10, 32)
+		if err == nil {
+			groups = append(groups, uint32(gval))
+		}
+	}
+
+	shell := userShell(u)
+
+	// Prefer the user's home directory as the working directory, but fall back
+	// to / if it does not exist (e.g. useradd was run without -m).
+	homeDir := u.HomeDir
+	if _, err := os.Stat(homeDir); err != nil {
+		homeDir = "/"
+	}
+
+	cmd := exec.Command(shell, "--login")
+	cmd.Env = []string{
+		"TERM=xterm-256color",
+		"HOME=" + u.HomeDir,
+		"USER=" + username,
+		"LOGNAME=" + username,
+		"SHELL=" + shell,
+		"PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin",
+	}
+	cmd.Dir = homeDir
+	cmd.SysProcAttr = &syscall.SysProcAttr{
+		Credential: &syscall.Credential{
+			Uid:    uint32(uid),
+			Gid:    uint32(gid),
+			Groups: groups,
+		},
+	}
+
+	ptmx, err := pty.Start(cmd)
+	if err != nil {
+		return nil, fmt.Errorf("pty start: %w", err)
+	}
+	return &PTYSession{ptmx: ptmx, cmd: cmd}, nil
+}

nativessh/server.go

@@ -0,0 +1,467 @@
+//go:build !windows
+
+package nativessh
+
+import (
+	"crypto/ed25519"
+	"crypto/rand"
+	"errors"
+	"fmt"
+	"io"
+	"net"
+	"os"
+	"os/exec"
+	"os/user"
+	"strconv"
+	"strings"
+	"sync"
+	"syscall"
+
+	"github.com/fosrl/newt/logger"
+	"golang.org/x/crypto/ssh"
+)
+
+// CredentialStore holds in-memory SSH credentials that can be updated at runtime.
+// It is safe for concurrent use.
+type CredentialStore struct {
+	mu         sync.RWMutex
+	caKey      ssh.PublicKey
+	principals map[string]map[string]struct{} // username -> set of allowed principals
+}
+
+// connMetaWithUser wraps ConnMetadata while overriding User() for cert checks.
+type connMetaWithUser struct {
+	ssh.ConnMetadata
+	user string
+}
+
+func (m connMetaWithUser) User() string { return m.user }
+
+// NewCredentialStore returns an empty, ready-to-use CredentialStore.
+func NewCredentialStore() *CredentialStore {
+	return &CredentialStore{
+		principals: make(map[string]map[string]struct{}),
+	}
+}
+
+// SetCAKey parses and stores the CA public key from authorized_keys-format data.
+func (s *CredentialStore) SetCAKey(authorizedKeyData string) error {
+	key, _, _, _, err := ssh.ParseAuthorizedKey([]byte(authorizedKeyData))
+	if err != nil {
+		return fmt.Errorf("parse CA key: %w", err)
+	}
+	s.mu.Lock()
+	s.caKey = key
+	s.mu.Unlock()
+	return nil
+}
+
+// AddPrincipals records username and niceId as allowed principals for username.
+// Both values are stored; either can appear in the certificate's ValidPrincipals
+// field to satisfy the standard cert-auth principal check.
+func (s *CredentialStore) AddPrincipals(username, niceId string) {
+	username = strings.TrimSpace(username)
+	niceId = strings.TrimSpace(niceId)
+	if username == "" {
+		return
+	}
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	if s.principals[username] == nil {
+		s.principals[username] = make(map[string]struct{})
+	}
+	s.principals[username][username] = struct{}{}
+	if niceId != "" {
+		s.principals[username][niceId] = struct{}{}
+	}
+}
+
+// get returns the CA key and the principal set for username under a read lock.
+func (s *CredentialStore) get(username string) (ssh.PublicKey, map[string]struct{}) {
+	s.mu.RLock()
+	defer s.mu.RUnlock()
+	return s.caKey, s.principals[username]
+}
+
+// ServerConfig holds configuration for the native SSH server.
+type ServerConfig struct {
+	// ListenAddr is the TCP address to listen on. Defaults to ":2222".
+	ListenAddr string
+	// Credentials provides in-memory CA key and per-user principals.
+	// Updates to the store are reflected immediately for new connections.
+	// If nil or the store has no CA key set, all connections are rejected.
+	Credentials *CredentialStore
+}
+
+// Server is a simple SSH server that authenticates clients via SSH certificate
+// auth only. Certificates must be signed by the configured CA and the
+// connecting username must appear in both the certificate's principal list and
+// the local principals file.
+type Server struct {
+	cfg ServerConfig
+}
+
+// NewServer creates a new Server. The ListenAddr defaults to ":2222" when empty.
+func NewServer(cfg ServerConfig) *Server {
+	if cfg.ListenAddr == "" {
+		cfg.ListenAddr = ":2222"
+	}
+	return &Server{cfg: cfg}
+}
+
+// buildSSHConfig builds the ssh.ServerConfig with multi-method authentication:
+//  1. Public key: host ~/.ssh/authorized_keys, then CA certificate.
+//  2. Password: system PAM stack (Linux only).
+func (s *Server) buildSSHConfig() (*ssh.ServerConfig, error) {
+	hostSigner, err := generateHostKey()
+	if err != nil {
+		return nil, fmt.Errorf("host key: %w", err)
+	}
+	cfg := &ssh.ServerConfig{
+		PublicKeyCallback: makePublicKeyCallback(s.cfg.Credentials),
+		PasswordCallback:  makePasswordCallback(),
+	}
+	cfg.AddHostKey(hostSigner)
+	return cfg, nil
+}
+
+// Serve accepts connections on ln and handles them. It returns when ln is
+// closed or a non-temporary Accept error occurs.
+func (s *Server) Serve(ln net.Listener) error {
+	sshCfg, err := s.buildSSHConfig()
+	if err != nil {
+		return err
+	}
+	logger.Debug("nativessh: server listening on %s", ln.Addr())
+	for {
+		conn, err := ln.Accept()
+		if err != nil {
+			return err
+		}
+		go s.handleConn(conn, sshCfg)
+	}
+}
+
+// ListenAndServe starts the SSH server on the host network and blocks until
+// the listener is closed.
+func (s *Server) ListenAndServe() error {
+	ln, err := net.Listen("tcp", s.cfg.ListenAddr)
+	if err != nil {
+		return fmt.Errorf("listen %s: %w", s.cfg.ListenAddr, err)
+	}
+	defer ln.Close()
+	return s.Serve(ln)
+}
+
+func (s *Server) handleConn(conn net.Conn, cfg *ssh.ServerConfig) {
+	defer conn.Close()
+	sshConn, chans, reqs, err := ssh.NewServerConn(conn, cfg)
+	if err != nil {
+		logger.Debug("nativessh: handshake failed from %s: %v", conn.RemoteAddr(), err)
+		return
+	}
+	defer sshConn.Close()
+	logger.Debug("nativessh: connection from %s user=%s", conn.RemoteAddr(), sshConn.User())
+
+	go ssh.DiscardRequests(reqs)
+
+	for newChan := range chans {
+		if newChan.ChannelType() != "session" {
+			_ = newChan.Reject(ssh.UnknownChannelType, "unknown channel type")
+			continue
+		}
+		ch, requests, err := newChan.Accept()
+		if err != nil {
+			logger.Debug("nativessh: channel accept error: %v", err)
+			return
+		}
+		go s.handleSession(ch, requests, sshConn.User())
+	}
+}
+
+// handleSession drives a single SSH session channel. It waits for a pty-req
+// followed by a shell request and then bridges the PTY to the channel.
+func (s *Server) handleSession(ch ssh.Channel, requests <-chan *ssh.Request, username string) {
+	defer ch.Close()
+
+	var (
+		sess    *PTYSession
+		started bool
+	)
+
+	for req := range requests {
+		switch req.Type {
+		case "pty-req":
+			var err error
+			if sess == nil {
+				sess, err = NewPTYSessionAs(username)
+				if err != nil {
+					logger.Debug("nativessh: PTY start error: %v", err)
+					if req.WantReply {
+						_ = req.Reply(false, nil)
+					}
+					return
+				}
+			}
+			cols, rows := parsePTYReq(req.Payload)
+			_ = sess.Resize(cols, rows)
+			if req.WantReply {
+				_ = req.Reply(true, nil)
+			}
+
+		case "shell":
+			if req.WantReply {
+				_ = req.Reply(true, nil)
+			}
+			if started || sess == nil {
+				continue
+			}
+			started = true
+			// PTY output → SSH channel.
+			go func() {
+				_, _ = io.Copy(ch, sess)
+				// Notify the client of the shell's exit status so it can
+				// disconnect cleanly instead of requiring a manual disconnect.
+				exitCode := sess.ExitCode()
+				exitStatusPayload := ssh.Marshal(struct{ Status uint32 }{uint32(exitCode)})
+				_, _ = ch.SendRequest("exit-status", false, exitStatusPayload)
+				_ = ch.CloseWrite()
+				sess.Close() //nolint:errcheck
+				// Close the channel so the ssh library closes the requests
+				// channel, which unblocks the for-range loop in handleSession
+				// and allows the deferred ch.Close() to run. Without this,
+				// handleSession blocks forever waiting for requests to drain.
+				_ = ch.Close()
+			}()
+			// SSH channel input → PTY stdin.
+			go func() {
+				_, _ = io.Copy(sess, ch)
+			}()
+
+		case "exec":
+			if req.WantReply {
+				_ = req.Reply(true, nil)
+			}
+			cmd := parseExecPayload(req.Payload)
+			go s.runExecAs(ch, username, cmd)
+			// Drain remaining requests; the goroutine owns the channel now.
+			go ssh.DiscardRequests(requests)
+			return
+
+		case "window-change":
+			if sess != nil {
+				cols, rows := parseWindowChange(req.Payload)
+				_ = sess.Resize(cols, rows)
+			}
+			if req.WantReply {
+				_ = req.Reply(true, nil)
+			}
+
+		default:
+			if req.WantReply {
+				_ = req.Reply(false, nil)
+			}
+		}
+	}
+
+	if sess != nil && !started {
+		sess.Close() //nolint:errcheck
+	}
+}
+
+// runExecAs runs command as username, wiring stdin/stdout/stderr directly to
+// the SSH channel (no PTY). This supports scp, rsync, and plain exec requests.
+func (s *Server) runExecAs(ch ssh.Channel, username, command string) {
+	defer ch.Close()
+
+	u, err := user.Lookup(username)
+	if err != nil {
+		logger.Debug("nativessh: exec user lookup %q: %v", username, err)
+		sendExitStatus(ch, 1)
+		return
+	}
+
+	uid, err := strconv.ParseUint(u.Uid, 10, 32)
+	if err != nil {
+		logger.Debug("nativessh: exec parse uid for %q: %v", username, err)
+		sendExitStatus(ch, 1)
+		return
+	}
+	gid, err := strconv.ParseUint(u.Gid, 10, 32)
+	if err != nil {
+		logger.Debug("nativessh: exec parse gid for %q: %v", username, err)
+		sendExitStatus(ch, 1)
+		return
+	}
+
+	groupIDs, _ := u.GroupIds()
+	var groups []uint32
+	for _, g := range groupIDs {
+		gval, err := strconv.ParseUint(g, 10, 32)
+		if err == nil {
+			groups = append(groups, uint32(gval))
+		}
+	}
+
+	homeDir := u.HomeDir
+	if _, err := os.Stat(homeDir); err != nil {
+		homeDir = "/"
+	}
+
+	cmd := exec.Command("/bin/sh", "-c", command)
+	cmd.Env = []string{
+		"HOME=" + u.HomeDir,
+		"USER=" + username,
+		"LOGNAME=" + username,
+		"PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin",
+	}
+	cmd.Dir = homeDir
+	cmd.SysProcAttr = &syscall.SysProcAttr{
+		Credential: &syscall.Credential{
+			Uid:    uint32(uid),
+			Gid:    uint32(gid),
+			Groups: groups,
+		},
+	}
+	cmd.Stdin = ch
+	cmd.Stdout = ch
+	cmd.Stderr = ch.Stderr()
+
+	exitCode := 0
+	if err := cmd.Run(); err != nil {
+		var exitErr *exec.ExitError
+		if errors.As(err, &exitErr) {
+			exitCode = exitErr.ExitCode()
+		} else {
+			exitCode = 1
+		}
+		logger.Debug("nativessh: exec %q as %q exited: %v", command, username, err)
+	}
+
+	sendExitStatus(ch, exitCode)
+	_ = ch.CloseWrite()
+}
+
+// sendExitStatus sends an SSH exit-status request on ch.
+func sendExitStatus(ch ssh.Channel, code int) {
+	payload := ssh.Marshal(struct{ Status uint32 }{uint32(code)})
+	_, _ = ch.SendRequest("exit-status", false, payload)
+}
+
+// parseExecPayload decodes the command string from an SSH exec request payload.
+func parseExecPayload(payload []byte) string {
+	var msg struct{ Command string }
+	if err := ssh.Unmarshal(payload, &msg); err != nil {
+		return ""
+	}
+	return msg.Command
+}
+
+// makePublicKeyCallback returns a PublicKeyCallback that tries, in order:
+//  1. Host authorized_keys  – matches any key in the OS user's
+//     ~/.ssh/authorized_keys file.
+//  2. CA certificate        – validates an SSH certificate signed by the
+//     configured CA and checks that the user appears in the principals map.
+//
+// store may be nil or empty; those paths are simply skipped.
+func makePublicKeyCallback(store *CredentialStore) func(ssh.ConnMetadata, ssh.PublicKey) (*ssh.Permissions, error) {
+	return func(meta ssh.ConnMetadata, key ssh.PublicKey) (*ssh.Permissions, error) {
+		// 1. Host authorized_keys.
+		if CheckAuthorizedKeys(meta.User(), key) {
+			logger.Debug("nativessh: authorized_keys auth for user %q", meta.User())
+			return &ssh.Permissions{}, nil
+		}
+
+		// 2. CA certificate.
+		if store != nil {
+			caKey, userPrincipals := store.get(meta.User())
+			if caKey != nil {
+				checker := &ssh.CertChecker{
+					IsUserAuthority: func(auth ssh.PublicKey) bool {
+						return ssh.FingerprintSHA256(auth) == ssh.FingerprintSHA256(caKey)
+					},
+				}
+
+				if len(userPrincipals) == 0 {
+					return nil, fmt.Errorf("user %q not in allowed principals list", meta.User())
+				}
+
+				var lastErr error
+				for principal := range userPrincipals {
+					perms, err := checker.Authenticate(connMetaWithUser{ConnMetadata: meta, user: principal}, key)
+					if err == nil {
+						logger.Debug("nativessh: CA cert auth for user %q principal=%q", meta.User(), principal)
+						return perms, nil
+					}
+					lastErr = err
+				}
+
+				if lastErr != nil {
+					logger.Debug("nativessh: CA cert rejected for user %q: %v", meta.User(), lastErr)
+				}
+			}
+		}
+
+		return nil, fmt.Errorf("public key not authorized for user %q", meta.User())
+	}
+}
+
+// makePasswordCallback returns a PasswordCallback that validates the supplied
+// password via the host OS PAM stack.  On non-Linux platforms this always
+// fails (see pam_other.go).
+func makePasswordCallback() func(ssh.ConnMetadata, []byte) (*ssh.Permissions, error) {
+	return func(meta ssh.ConnMetadata, password []byte) (*ssh.Permissions, error) {
+		if err := VerifySystemPassword(meta.User(), string(password)); err != nil {
+			// Return a generic message to the client; log the real reason.
+			logger.Debug("nativessh: password auth failed for user %q: %v", meta.User(), err)
+			return nil, fmt.Errorf("permission denied")
+		}
+		logger.Debug("nativessh: password auth for user %q", meta.User())
+		return &ssh.Permissions{}, nil
+	}
+}
+
+// generateHostKey generates a fresh ephemeral Ed25519 host key in memory.
+// A new key is created on every server start; nothing is written to disk.
+func generateHostKey() (ssh.Signer, error) {
+	_, priv, err := ed25519.GenerateKey(rand.Reader)
+	if err != nil {
+		return nil, fmt.Errorf("generate host key: %w", err)
+	}
+	logger.Debug("nativessh: generated ephemeral Ed25519 host key")
+	return ssh.NewSignerFromKey(priv)
+}
+
+// ptyRequestMsg mirrors the SSH wire format for pty-req (RFC 4254 §6.2).
+type ptyRequestMsg struct {
+	Term     string
+	Columns  uint32
+	Rows     uint32
+	Width    uint32
+	Height   uint32
+	Modelist string
+}
+
+func parsePTYReq(payload []byte) (cols, rows uint16) {
+	var req ptyRequestMsg
+	if err := ssh.Unmarshal(payload, &req); err != nil {
+		return 80, 24
+	}
+	return uint16(req.Columns), uint16(req.Rows)
+}
+
+// windowChangeMsg mirrors the SSH wire format for window-change (RFC 4254 §6.7).
+type windowChangeMsg struct {
+	Columns uint32
+	Rows    uint32
+	Width   uint32
+	Height  uint32
+}
+
+func parseWindowChange(payload []byte) (cols, rows uint16) {
+	var msg windowChangeMsg
+	if err := ssh.Unmarshal(payload, &msg); err != nil {
+		return 80, 24
+	}
+	return uint16(msg.Columns), uint16(msg.Rows)
+}

nativessh/server_windows.go

@@ -0,0 +1,64 @@
+//go:build windows
+
+package nativessh
+
+import (
+	"errors"
+	"log"
+	"net"
+	"sync"
+
+	"golang.org/x/crypto/ssh"
+)
+
+// CredentialStore is a stub on Windows. Native SSH is not supported on Windows.
+type CredentialStore struct {
+	mu         sync.RWMutex
+	principals map[string]map[string]struct{}
+}
+
+// NewCredentialStore returns an empty CredentialStore stub.
+// Native SSH is not supported on Windows; a warning is logged.
+func NewCredentialStore() *CredentialStore {
+	log.Println("WARNING: native SSH is not supported on Windows and will be disabled")
+	return &CredentialStore{
+		principals: make(map[string]map[string]struct{}),
+	}
+}
+
+// SetCAKey is a no-op stub on Windows.
+func (s *CredentialStore) SetCAKey(_ string) error {
+	return errors.New("native SSH not supported on Windows")
+}
+
+// AddPrincipals is a no-op stub on Windows.
+func (s *CredentialStore) AddPrincipals(_, _ string) {}
+
+// get returns nil CA key and empty principals on Windows.
+func (s *CredentialStore) get(_ string) (ssh.PublicKey, map[string]struct{}) {
+	return nil, nil
+}
+
+// ServerConfig holds configuration for the native SSH server (stub on Windows).
+type ServerConfig struct {
+	ListenAddr  string
+	Credentials *CredentialStore
+}
+
+// Server is a stub on Windows.
+type Server struct{}
+
+// NewServer returns a stub Server and logs a warning.
+func NewServer(cfg ServerConfig) *Server {
+	return &Server{}
+}
+
+// ListenAndServe always returns an error on Windows.
+func (s *Server) ListenAndServe() error {
+	return errors.New("native SSH not supported on Windows")
+}
+
+// Serve always returns an error on Windows.
+func (s *Server) Serve(_ net.Listener) error {
+	return errors.New("native SSH not supported on Windows")
+}

netstack2/access_log.go

@@ -0,0 +1,514 @@
+package netstack2
+
+import (
+	"bytes"
+	"compress/zlib"
+	"crypto/rand"
+	"encoding/base64"
+	"encoding/hex"
+	"encoding/json"
+	"net"
+	"sort"
+	"sync"
+	"time"
+
+	"github.com/fosrl/newt/logger"
+)
+
+const (
+	// flushInterval is how often the access logger flushes completed sessions to the server
+	flushInterval = 60 * time.Second
+
+	// maxBufferedSessions is the max number of completed sessions to buffer before forcing a flush
+	maxBufferedSessions = 100
+
+	// sessionGapThreshold is the maximum gap between the end of one connection
+	// and the start of the next for them to be considered part of the same session.
+	// If the gap exceeds this, a new consolidated session is created.
+	sessionGapThreshold = 5 * time.Second
+
+	// minConnectionsToConsolidate is the minimum number of connections in a group
+	// before we bother consolidating. Groups smaller than this are sent as-is.
+	minConnectionsToConsolidate = 2
+)
+
+// SendFunc is a callback that sends compressed access log data to the server.
+// The data is a base64-encoded zlib-compressed JSON array of AccessSession objects.
+type SendFunc func(data string) error
+
+// AccessSession represents a tracked access session through the proxy
+type AccessSession struct {
+	SessionID       string    `json:"sessionId"`
+	ResourceID      int       `json:"resourceId"`
+	SourceAddr      string    `json:"sourceAddr"`
+	DestAddr        string    `json:"destAddr"`
+	Protocol        string    `json:"protocol"`
+	StartedAt       time.Time `json:"startedAt"`
+	EndedAt         time.Time `json:"endedAt,omitempty"`
+	BytesTx         int64     `json:"bytesTx"`
+	BytesRx         int64     `json:"bytesRx"`
+	ConnectionCount int       `json:"connectionCount,omitempty"` // number of raw connections merged into this session (0 or 1 = single)
+}
+
+// udpSessionKey identifies a unique UDP "session" by src -> dst
+type udpSessionKey struct {
+	srcAddr  string
+	dstAddr  string
+	protocol string
+}
+
+// consolidationKey groups connections that may be part of the same logical session.
+// Source port is intentionally excluded so that many ephemeral-port connections
+// from the same source IP to the same destination are grouped together.
+type consolidationKey struct {
+	sourceIP   string // IP only, no port
+	destAddr   string // full host:port of the destination
+	protocol   string
+	resourceID int
+}
+
+// AccessLogger tracks access sessions for resources and periodically
+// flushes completed sessions to the server via a configurable SendFunc.
+type AccessLogger struct {
+	mu                sync.Mutex
+	sessions          map[string]*AccessSession       // active sessions: sessionID -> session
+	udpSessions       map[udpSessionKey]*AccessSession // active UDP sessions for dedup
+	completedSessions []*AccessSession                 // completed sessions waiting to be flushed
+	udpTimeout        time.Duration
+	sendFn            SendFunc
+	stopCh            chan struct{}
+	flushDone         chan struct{} // closed after the flush goroutine exits
+}
+
+// NewAccessLogger creates a new access logger.
+// udpTimeout controls how long a UDP session is kept alive without traffic before being ended.
+func NewAccessLogger(udpTimeout time.Duration) *AccessLogger {
+	al := &AccessLogger{
+		sessions:          make(map[string]*AccessSession),
+		udpSessions:       make(map[udpSessionKey]*AccessSession),
+		completedSessions: make([]*AccessSession, 0),
+		udpTimeout:        udpTimeout,
+		stopCh:            make(chan struct{}),
+		flushDone:         make(chan struct{}),
+	}
+	go al.backgroundLoop()
+	return al
+}
+
+// SetSendFunc sets the callback used to send compressed access log batches
+// to the server. This can be called after construction once the websocket
+// client is available.
+func (al *AccessLogger) SetSendFunc(fn SendFunc) {
+	al.mu.Lock()
+	defer al.mu.Unlock()
+	al.sendFn = fn
+}
+
+// generateSessionID creates a random session identifier
+func generateSessionID() string {
+	b := make([]byte, 8)
+	rand.Read(b)
+	return hex.EncodeToString(b)
+}
+
+// StartTCPSession logs the start of a TCP session and returns a session ID.
+func (al *AccessLogger) StartTCPSession(resourceID int, srcAddr, dstAddr string) string {
+	sessionID := generateSessionID()
+	now := time.Now()
+
+	session := &AccessSession{
+		SessionID:  sessionID,
+		ResourceID: resourceID,
+		SourceAddr: srcAddr,
+		DestAddr:   dstAddr,
+		Protocol:   "tcp",
+		StartedAt:  now,
+	}
+
+	al.mu.Lock()
+	al.sessions[sessionID] = session
+	al.mu.Unlock()
+
+	logger.Info("ACCESS START session=%s resource=%d proto=tcp src=%s dst=%s time=%s",
+		sessionID, resourceID, srcAddr, dstAddr, now.Format(time.RFC3339))
+
+	return sessionID
+}
+
+// EndTCPSession logs the end of a TCP session and queues it for sending.
+func (al *AccessLogger) EndTCPSession(sessionID string) {
+	now := time.Now()
+
+	al.mu.Lock()
+	session, ok := al.sessions[sessionID]
+	if ok {
+		session.EndedAt = now
+		delete(al.sessions, sessionID)
+		al.completedSessions = append(al.completedSessions, session)
+	}
+	shouldFlush := len(al.completedSessions) >= maxBufferedSessions
+	al.mu.Unlock()
+
+	if ok {
+		duration := now.Sub(session.StartedAt)
+		logger.Info("ACCESS END session=%s resource=%d proto=tcp src=%s dst=%s started=%s ended=%s duration=%s",
+			sessionID, session.ResourceID, session.SourceAddr, session.DestAddr,
+			session.StartedAt.Format(time.RFC3339), now.Format(time.RFC3339), duration)
+	}
+
+	if shouldFlush {
+		al.flush()
+	}
+}
+
+// TrackUDPSession starts or returns an existing UDP session. Returns the session ID.
+func (al *AccessLogger) TrackUDPSession(resourceID int, srcAddr, dstAddr string) string {
+	key := udpSessionKey{
+		srcAddr:  srcAddr,
+		dstAddr:  dstAddr,
+		protocol: "udp",
+	}
+
+	al.mu.Lock()
+	defer al.mu.Unlock()
+
+	if existing, ok := al.udpSessions[key]; ok {
+		return existing.SessionID
+	}
+
+	sessionID := generateSessionID()
+	now := time.Now()
+
+	session := &AccessSession{
+		SessionID:  sessionID,
+		ResourceID: resourceID,
+		SourceAddr: srcAddr,
+		DestAddr:   dstAddr,
+		Protocol:   "udp",
+		StartedAt:  now,
+	}
+
+	al.sessions[sessionID] = session
+	al.udpSessions[key] = session
+
+	logger.Info("ACCESS START session=%s resource=%d proto=udp src=%s dst=%s time=%s",
+		sessionID, resourceID, srcAddr, dstAddr, now.Format(time.RFC3339))
+
+	return sessionID
+}
+
+// EndUDPSession ends a UDP session and queues it for sending.
+func (al *AccessLogger) EndUDPSession(sessionID string) {
+	now := time.Now()
+
+	al.mu.Lock()
+	session, ok := al.sessions[sessionID]
+	if ok {
+		session.EndedAt = now
+		delete(al.sessions, sessionID)
+		key := udpSessionKey{
+			srcAddr:  session.SourceAddr,
+			dstAddr:  session.DestAddr,
+			protocol: "udp",
+		}
+		delete(al.udpSessions, key)
+		al.completedSessions = append(al.completedSessions, session)
+	}
+	shouldFlush := len(al.completedSessions) >= maxBufferedSessions
+	al.mu.Unlock()
+
+	if ok {
+		duration := now.Sub(session.StartedAt)
+		logger.Info("ACCESS END session=%s resource=%d proto=udp src=%s dst=%s started=%s ended=%s duration=%s",
+			sessionID, session.ResourceID, session.SourceAddr, session.DestAddr,
+			session.StartedAt.Format(time.RFC3339), now.Format(time.RFC3339), duration)
+	}
+
+	if shouldFlush {
+		al.flush()
+	}
+}
+
+// backgroundLoop handles periodic flushing and stale session reaping.
+func (al *AccessLogger) backgroundLoop() {
+	defer close(al.flushDone)
+
+	flushTicker := time.NewTicker(flushInterval)
+	defer flushTicker.Stop()
+
+	reapTicker := time.NewTicker(30 * time.Second)
+	defer reapTicker.Stop()
+
+	for {
+		select {
+		case <-al.stopCh:
+			return
+		case <-flushTicker.C:
+			al.flush()
+		case <-reapTicker.C:
+			al.reapStaleSessions()
+		}
+	}
+}
+
+// reapStaleSessions cleans up UDP sessions that were not properly ended.
+func (al *AccessLogger) reapStaleSessions() {
+	al.mu.Lock()
+	defer al.mu.Unlock()
+
+	staleThreshold := time.Now().Add(-5 * time.Minute)
+
+	for key, session := range al.udpSessions {
+		if session.StartedAt.Before(staleThreshold) && session.EndedAt.IsZero() {
+			now := time.Now()
+			session.EndedAt = now
+			duration := now.Sub(session.StartedAt)
+			logger.Info("ACCESS END (reaped) session=%s resource=%d proto=udp src=%s dst=%s started=%s ended=%s duration=%s",
+				session.SessionID, session.ResourceID, session.SourceAddr, session.DestAddr,
+				session.StartedAt.Format(time.RFC3339), now.Format(time.RFC3339), duration)
+			al.completedSessions = append(al.completedSessions, session)
+			delete(al.sessions, session.SessionID)
+			delete(al.udpSessions, key)
+		}
+	}
+}
+
+// extractIP strips the port from an address string and returns just the IP.
+// If the address has no port component it is returned as-is.
+func extractIP(addr string) string {
+	host, _, err := net.SplitHostPort(addr)
+	if err != nil {
+		// Might already be a bare IP
+		return addr
+	}
+	return host
+}
+
+// consolidateSessions takes a slice of completed sessions and merges bursts of
+// short-lived connections from the same source IP to the same destination into
+// single higher-level session entries.
+//
+// The algorithm:
+//  1. Group sessions by (sourceIP, destAddr, protocol, resourceID).
+//  2. Within each group, sort by StartedAt.
+//  3. Walk through the sorted list and merge consecutive sessions whose gap
+//     (previous EndedAt → next StartedAt) is ≤ sessionGapThreshold.
+//  4. For merged sessions the earliest StartedAt and latest EndedAt are kept,
+//     bytes are summed, and ConnectionCount records how many raw connections
+//     were folded in. If the merged connections used more than one source port,
+//     SourceAddr is set to just the IP (port omitted).
+//  5. Groups with fewer than minConnectionsToConsolidate members are passed
+//     through unmodified.
+func consolidateSessions(sessions []*AccessSession) []*AccessSession {
+	if len(sessions) <= 1 {
+		return sessions
+	}
+
+	// Group sessions by consolidation key
+	groups := make(map[consolidationKey][]*AccessSession)
+	for _, s := range sessions {
+		key := consolidationKey{
+			sourceIP:   extractIP(s.SourceAddr),
+			destAddr:   s.DestAddr,
+			protocol:   s.Protocol,
+			resourceID: s.ResourceID,
+		}
+		groups[key] = append(groups[key], s)
+	}
+
+	result := make([]*AccessSession, 0, len(sessions))
+
+	for key, group := range groups {
+		// Small groups don't need consolidation
+		if len(group) < minConnectionsToConsolidate {
+			result = append(result, group...)
+			continue
+		}
+
+		// Sort the group by start time so we can detect gaps
+		sort.Slice(group, func(i, j int) bool {
+			return group[i].StartedAt.Before(group[j].StartedAt)
+		})
+
+		// Walk through and merge runs that are within the gap threshold
+		var merged []*AccessSession
+		cur := cloneSession(group[0])
+		cur.ConnectionCount = 1
+		sourcePorts := make(map[string]struct{})
+		sourcePorts[cur.SourceAddr] = struct{}{}
+
+		for i := 1; i < len(group); i++ {
+			s := group[i]
+
+			// Determine the gap: from the latest end time we've seen so far to the
+			// start of the next connection.
+			gapRef := cur.EndedAt
+			if gapRef.IsZero() {
+				gapRef = cur.StartedAt
+			}
+			gap := s.StartedAt.Sub(gapRef)
+
+			if gap <= sessionGapThreshold {
+				// Merge into the current consolidated session
+				cur.ConnectionCount++
+				cur.BytesTx += s.BytesTx
+				cur.BytesRx += s.BytesRx
+				sourcePorts[s.SourceAddr] = struct{}{}
+
+				// Extend EndedAt to the latest time
+				endTime := s.EndedAt
+				if endTime.IsZero() {
+					endTime = s.StartedAt
+				}
+				if endTime.After(cur.EndedAt) {
+					cur.EndedAt = endTime
+				}
+			} else {
+				// Gap exceeded — finalize the current session and start a new one
+				finalizeMergedSourceAddr(cur, key.sourceIP, sourcePorts)
+				merged = append(merged, cur)
+
+				cur = cloneSession(s)
+				cur.ConnectionCount = 1
+				sourcePorts = make(map[string]struct{})
+				sourcePorts[s.SourceAddr] = struct{}{}
+			}
+		}
+
+		// Finalize the last accumulated session
+		finalizeMergedSourceAddr(cur, key.sourceIP, sourcePorts)
+		merged = append(merged, cur)
+
+		result = append(result, merged...)
+	}
+
+	return result
+}
+
+// cloneSession creates a shallow copy of an AccessSession.
+func cloneSession(s *AccessSession) *AccessSession {
+	cp := *s
+	return &cp
+}
+
+// finalizeMergedSourceAddr sets the SourceAddr on a consolidated session.
+// If multiple distinct source addresses (ports) were seen, the port is
+// stripped and only the IP is kept so the log isn't misleading.
+func finalizeMergedSourceAddr(s *AccessSession, sourceIP string, ports map[string]struct{}) {
+	if len(ports) > 1 {
+		// Multiple source ports — just report the IP
+		s.SourceAddr = sourceIP
+	}
+	// Otherwise keep the original SourceAddr which already has ip:port
+}
+
+// flush drains the completed sessions buffer, consolidates bursts of
+// short-lived connections, compresses with zlib, and sends via the SendFunc.
+func (al *AccessLogger) flush() {
+	al.mu.Lock()
+	if len(al.completedSessions) == 0 {
+		al.mu.Unlock()
+		return
+	}
+	batch := al.completedSessions
+	al.completedSessions = make([]*AccessSession, 0)
+	sendFn := al.sendFn
+	al.mu.Unlock()
+
+	if sendFn == nil {
+		logger.Debug("Access logger: no send function configured, discarding %d sessions", len(batch))
+		return
+	}
+
+	// Consolidate bursts of short-lived connections into higher-level sessions
+	originalCount := len(batch)
+	batch = consolidateSessions(batch)
+	if len(batch) != originalCount {
+		logger.Info("Access logger: consolidated %d raw connections into %d sessions", originalCount, len(batch))
+	}
+
+	compressed, err := compressSessions(batch)
+	if err != nil {
+		logger.Error("Access logger: failed to compress %d sessions: %v", len(batch), err)
+		return
+	}
+
+	if err := sendFn(compressed); err != nil {
+		logger.Error("Access logger: failed to send %d sessions: %v", len(batch), err)
+		// Re-queue the batch so we don't lose data
+		al.mu.Lock()
+		al.completedSessions = append(batch, al.completedSessions...)
+		// Cap re-queued data to prevent unbounded growth if server is unreachable
+		if len(al.completedSessions) > maxBufferedSessions*5 {
+			dropped := len(al.completedSessions) - maxBufferedSessions*5
+			al.completedSessions = al.completedSessions[:maxBufferedSessions*5]
+			logger.Warn("Access logger: buffer overflow, dropped %d oldest sessions", dropped)
+		}
+		al.mu.Unlock()
+		return
+	}
+
+	logger.Info("Access logger: sent %d sessions to server", len(batch))
+}
+
+// compressSessions JSON-encodes the sessions, compresses with zlib, and returns
+// a base64-encoded string suitable for embedding in a JSON message.
+func compressSessions(sessions []*AccessSession) (string, error) {
+	jsonData, err := json.Marshal(sessions)
+	if err != nil {
+		return "", err
+	}
+
+	var buf bytes.Buffer
+	w, err := zlib.NewWriterLevel(&buf, zlib.BestCompression)
+	if err != nil {
+		return "", err
+	}
+	if _, err := w.Write(jsonData); err != nil {
+		w.Close()
+		return "", err
+	}
+	if err := w.Close(); err != nil {
+		return "", err
+	}
+
+	return base64.StdEncoding.EncodeToString(buf.Bytes()), nil
+}
+
+// Close shuts down the background loop, ends all active sessions,
+// and performs one final flush to send everything to the server.
+func (al *AccessLogger) Close() {
+	// Signal the background loop to stop
+	select {
+	case <-al.stopCh:
+		// Already closed
+		return
+	default:
+		close(al.stopCh)
+	}
+
+	// Wait for the background loop to exit so we don't race on flush
+	<-al.flushDone
+
+	al.mu.Lock()
+	now := time.Now()
+
+	// End all active sessions and move them to the completed buffer
+	for _, session := range al.sessions {
+		if session.EndedAt.IsZero() {
+			session.EndedAt = now
+			duration := now.Sub(session.StartedAt)
+			logger.Info("ACCESS END (shutdown) session=%s resource=%d proto=%s src=%s dst=%s started=%s ended=%s duration=%s",
+				session.SessionID, session.ResourceID, session.Protocol, session.SourceAddr, session.DestAddr,
+				session.StartedAt.Format(time.RFC3339), now.Format(time.RFC3339), duration)
+			al.completedSessions = append(al.completedSessions, session)
+		}
+	}
+
+	al.sessions = make(map[string]*AccessSession)
+	al.udpSessions = make(map[udpSessionKey]*AccessSession)
+	al.mu.Unlock()
+
+	// Final flush to send all remaining sessions to the server
+	al.flush()
+}

netstack2/access_log_test.go

@@ -0,0 +1,811 @@
+package netstack2
+
+import (
+	"testing"
+	"time"
+)
+
+func TestExtractIP(t *testing.T) {
+	tests := []struct {
+		name     string
+		addr     string
+		expected string
+	}{
+		{"ipv4 with port", "192.168.1.1:12345", "192.168.1.1"},
+		{"ipv4 without port", "192.168.1.1", "192.168.1.1"},
+		{"ipv6 with port", "[::1]:12345", "::1"},
+		{"ipv6 without port", "::1", "::1"},
+		{"empty string", "", ""},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			result := extractIP(tt.addr)
+			if result != tt.expected {
+				t.Errorf("extractIP(%q) = %q, want %q", tt.addr, result, tt.expected)
+			}
+		})
+	}
+}
+
+func TestConsolidateSessions_Empty(t *testing.T) {
+	result := consolidateSessions(nil)
+	if result != nil {
+		t.Errorf("expected nil, got %v", result)
+	}
+
+	result = consolidateSessions([]*AccessSession{})
+	if len(result) != 0 {
+		t.Errorf("expected empty slice, got %d items", len(result))
+	}
+}
+
+func TestConsolidateSessions_SingleSession(t *testing.T) {
+	now := time.Now()
+	sessions := []*AccessSession{
+		{
+			SessionID:  "abc123",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5000",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now,
+			EndedAt:    now.Add(1 * time.Second),
+		},
+	}
+
+	result := consolidateSessions(sessions)
+	if len(result) != 1 {
+		t.Fatalf("expected 1 session, got %d", len(result))
+	}
+	if result[0].SourceAddr != "10.0.0.1:5000" {
+		t.Errorf("expected source addr preserved, got %q", result[0].SourceAddr)
+	}
+}
+
+func TestConsolidateSessions_MergesBurstFromSameSourceIP(t *testing.T) {
+	now := time.Now()
+	sessions := []*AccessSession{
+		{
+			SessionID:  "s1",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5000",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now,
+			EndedAt:    now.Add(100 * time.Millisecond),
+			BytesTx:    100,
+			BytesRx:    200,
+		},
+		{
+			SessionID:  "s2",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5001",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now.Add(200 * time.Millisecond),
+			EndedAt:    now.Add(300 * time.Millisecond),
+			BytesTx:    150,
+			BytesRx:    250,
+		},
+		{
+			SessionID:  "s3",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5002",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now.Add(400 * time.Millisecond),
+			EndedAt:    now.Add(500 * time.Millisecond),
+			BytesTx:    50,
+			BytesRx:    75,
+		},
+	}
+
+	result := consolidateSessions(sessions)
+	if len(result) != 1 {
+		t.Fatalf("expected 1 consolidated session, got %d", len(result))
+	}
+
+	s := result[0]
+	if s.ConnectionCount != 3 {
+		t.Errorf("expected ConnectionCount=3, got %d", s.ConnectionCount)
+	}
+	if s.SourceAddr != "10.0.0.1" {
+		t.Errorf("expected source addr to be IP only (multiple ports), got %q", s.SourceAddr)
+	}
+	if s.DestAddr != "192.168.1.100:443" {
+		t.Errorf("expected dest addr preserved, got %q", s.DestAddr)
+	}
+	if s.StartedAt != now {
+		t.Errorf("expected StartedAt to be earliest time")
+	}
+	if s.EndedAt != now.Add(500*time.Millisecond) {
+		t.Errorf("expected EndedAt to be latest time")
+	}
+	expectedTx := int64(300)
+	expectedRx := int64(525)
+	if s.BytesTx != expectedTx {
+		t.Errorf("expected BytesTx=%d, got %d", expectedTx, s.BytesTx)
+	}
+	if s.BytesRx != expectedRx {
+		t.Errorf("expected BytesRx=%d, got %d", expectedRx, s.BytesRx)
+	}
+}
+
+func TestConsolidateSessions_SameSourcePortPreserved(t *testing.T) {
+	now := time.Now()
+	sessions := []*AccessSession{
+		{
+			SessionID:  "s1",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5000",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now,
+			EndedAt:    now.Add(100 * time.Millisecond),
+		},
+		{
+			SessionID:  "s2",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5000",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now.Add(200 * time.Millisecond),
+			EndedAt:    now.Add(300 * time.Millisecond),
+		},
+	}
+
+	result := consolidateSessions(sessions)
+	if len(result) != 1 {
+		t.Fatalf("expected 1 session, got %d", len(result))
+	}
+	if result[0].SourceAddr != "10.0.0.1:5000" {
+		t.Errorf("expected source addr with port preserved when all ports are the same, got %q", result[0].SourceAddr)
+	}
+	if result[0].ConnectionCount != 2 {
+		t.Errorf("expected ConnectionCount=2, got %d", result[0].ConnectionCount)
+	}
+}
+
+func TestConsolidateSessions_GapSplitsSessions(t *testing.T) {
+	now := time.Now()
+
+	// First burst
+	sessions := []*AccessSession{
+		{
+			SessionID:  "s1",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5000",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now,
+			EndedAt:    now.Add(100 * time.Millisecond),
+		},
+		{
+			SessionID:  "s2",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5001",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now.Add(200 * time.Millisecond),
+			EndedAt:    now.Add(300 * time.Millisecond),
+		},
+		// Big gap here (10 seconds)
+		{
+			SessionID:  "s3",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5002",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now.Add(10 * time.Second),
+			EndedAt:    now.Add(10*time.Second + 100*time.Millisecond),
+		},
+		{
+			SessionID:  "s4",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5003",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now.Add(10*time.Second + 200*time.Millisecond),
+			EndedAt:    now.Add(10*time.Second + 300*time.Millisecond),
+		},
+	}
+
+	result := consolidateSessions(sessions)
+	if len(result) != 2 {
+		t.Fatalf("expected 2 consolidated sessions (gap split), got %d", len(result))
+	}
+
+	// Find the sessions by their start time
+	var first, second *AccessSession
+	for _, s := range result {
+		if s.StartedAt.Equal(now) {
+			first = s
+		} else {
+			second = s
+		}
+	}
+
+	if first == nil || second == nil {
+		t.Fatal("could not find both consolidated sessions")
+	}
+
+	if first.ConnectionCount != 2 {
+		t.Errorf("first burst: expected ConnectionCount=2, got %d", first.ConnectionCount)
+	}
+	if second.ConnectionCount != 2 {
+		t.Errorf("second burst: expected ConnectionCount=2, got %d", second.ConnectionCount)
+	}
+}
+
+func TestConsolidateSessions_DifferentDestinationsNotMerged(t *testing.T) {
+	now := time.Now()
+	sessions := []*AccessSession{
+		{
+			SessionID:  "s1",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5000",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now,
+			EndedAt:    now.Add(100 * time.Millisecond),
+		},
+		{
+			SessionID:  "s2",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5001",
+			DestAddr:   "192.168.1.100:8080",
+			Protocol:   "tcp",
+			StartedAt:  now.Add(200 * time.Millisecond),
+			EndedAt:    now.Add(300 * time.Millisecond),
+		},
+	}
+
+	result := consolidateSessions(sessions)
+	// Each goes to a different dest port so they should not be merged
+	if len(result) != 2 {
+		t.Fatalf("expected 2 sessions (different destinations), got %d", len(result))
+	}
+}
+
+func TestConsolidateSessions_DifferentProtocolsNotMerged(t *testing.T) {
+	now := time.Now()
+	sessions := []*AccessSession{
+		{
+			SessionID:  "s1",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5000",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now,
+			EndedAt:    now.Add(100 * time.Millisecond),
+		},
+		{
+			SessionID:  "s2",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5001",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "udp",
+			StartedAt:  now.Add(200 * time.Millisecond),
+			EndedAt:    now.Add(300 * time.Millisecond),
+		},
+	}
+
+	result := consolidateSessions(sessions)
+	if len(result) != 2 {
+		t.Fatalf("expected 2 sessions (different protocols), got %d", len(result))
+	}
+}
+
+func TestConsolidateSessions_DifferentResourceIDsNotMerged(t *testing.T) {
+	now := time.Now()
+	sessions := []*AccessSession{
+		{
+			SessionID:  "s1",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5000",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now,
+			EndedAt:    now.Add(100 * time.Millisecond),
+		},
+		{
+			SessionID:  "s2",
+			ResourceID: 2,
+			SourceAddr: "10.0.0.1:5001",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now.Add(200 * time.Millisecond),
+			EndedAt:    now.Add(300 * time.Millisecond),
+		},
+	}
+
+	result := consolidateSessions(sessions)
+	if len(result) != 2 {
+		t.Fatalf("expected 2 sessions (different resource IDs), got %d", len(result))
+	}
+}
+
+func TestConsolidateSessions_DifferentSourceIPsNotMerged(t *testing.T) {
+	now := time.Now()
+	sessions := []*AccessSession{
+		{
+			SessionID:  "s1",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5000",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now,
+			EndedAt:    now.Add(100 * time.Millisecond),
+		},
+		{
+			SessionID:  "s2",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.2:5001",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now.Add(200 * time.Millisecond),
+			EndedAt:    now.Add(300 * time.Millisecond),
+		},
+	}
+
+	result := consolidateSessions(sessions)
+	if len(result) != 2 {
+		t.Fatalf("expected 2 sessions (different source IPs), got %d", len(result))
+	}
+}
+
+func TestConsolidateSessions_OutOfOrderInput(t *testing.T) {
+	now := time.Now()
+	// Provide sessions out of chronological order to verify sorting
+	sessions := []*AccessSession{
+		{
+			SessionID:  "s3",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5002",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now.Add(400 * time.Millisecond),
+			EndedAt:    now.Add(500 * time.Millisecond),
+			BytesTx:    30,
+		},
+		{
+			SessionID:  "s1",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5000",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now,
+			EndedAt:    now.Add(100 * time.Millisecond),
+			BytesTx:    10,
+		},
+		{
+			SessionID:  "s2",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5001",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now.Add(200 * time.Millisecond),
+			EndedAt:    now.Add(300 * time.Millisecond),
+			BytesTx:    20,
+		},
+	}
+
+	result := consolidateSessions(sessions)
+	if len(result) != 1 {
+		t.Fatalf("expected 1 consolidated session, got %d", len(result))
+	}
+
+	s := result[0]
+	if s.ConnectionCount != 3 {
+		t.Errorf("expected ConnectionCount=3, got %d", s.ConnectionCount)
+	}
+	if s.StartedAt != now {
+		t.Errorf("expected StartedAt to be earliest time")
+	}
+	if s.EndedAt != now.Add(500*time.Millisecond) {
+		t.Errorf("expected EndedAt to be latest time")
+	}
+	if s.BytesTx != 60 {
+		t.Errorf("expected BytesTx=60, got %d", s.BytesTx)
+	}
+}
+
+func TestConsolidateSessions_ExactlyAtGapThreshold(t *testing.T) {
+	now := time.Now()
+	sessions := []*AccessSession{
+		{
+			SessionID:  "s1",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5000",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now,
+			EndedAt:    now.Add(100 * time.Millisecond),
+		},
+		{
+			// Starts exactly sessionGapThreshold after s1 ends — should still merge
+			SessionID:  "s2",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5001",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now.Add(100*time.Millisecond + sessionGapThreshold),
+			EndedAt:    now.Add(100*time.Millisecond + sessionGapThreshold + 50*time.Millisecond),
+		},
+	}
+
+	result := consolidateSessions(sessions)
+	if len(result) != 1 {
+		t.Fatalf("expected 1 session (gap exactly at threshold merges), got %d", len(result))
+	}
+	if result[0].ConnectionCount != 2 {
+		t.Errorf("expected ConnectionCount=2, got %d", result[0].ConnectionCount)
+	}
+}
+
+func TestConsolidateSessions_JustOverGapThreshold(t *testing.T) {
+	now := time.Now()
+	sessions := []*AccessSession{
+		{
+			SessionID:  "s1",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5000",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now,
+			EndedAt:    now.Add(100 * time.Millisecond),
+		},
+		{
+			// Starts 1ms over the gap threshold after s1 ends — should split
+			SessionID:  "s2",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5001",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now.Add(100*time.Millisecond + sessionGapThreshold + 1*time.Millisecond),
+			EndedAt:    now.Add(100*time.Millisecond + sessionGapThreshold + 50*time.Millisecond),
+		},
+	}
+
+	result := consolidateSessions(sessions)
+	if len(result) != 2 {
+		t.Fatalf("expected 2 sessions (gap just over threshold splits), got %d", len(result))
+	}
+}
+
+func TestConsolidateSessions_UDPSessions(t *testing.T) {
+	now := time.Now()
+	sessions := []*AccessSession{
+		{
+			SessionID:  "u1",
+			ResourceID: 5,
+			SourceAddr: "10.0.0.1:6000",
+			DestAddr:   "192.168.1.100:53",
+			Protocol:   "udp",
+			StartedAt:  now,
+			EndedAt:    now.Add(50 * time.Millisecond),
+			BytesTx:    64,
+			BytesRx:    512,
+		},
+		{
+			SessionID:  "u2",
+			ResourceID: 5,
+			SourceAddr: "10.0.0.1:6001",
+			DestAddr:   "192.168.1.100:53",
+			Protocol:   "udp",
+			StartedAt:  now.Add(100 * time.Millisecond),
+			EndedAt:    now.Add(150 * time.Millisecond),
+			BytesTx:    64,
+			BytesRx:    256,
+		},
+		{
+			SessionID:  "u3",
+			ResourceID: 5,
+			SourceAddr: "10.0.0.1:6002",
+			DestAddr:   "192.168.1.100:53",
+			Protocol:   "udp",
+			StartedAt:  now.Add(200 * time.Millisecond),
+			EndedAt:    now.Add(250 * time.Millisecond),
+			BytesTx:    64,
+			BytesRx:    128,
+		},
+	}
+
+	result := consolidateSessions(sessions)
+	if len(result) != 1 {
+		t.Fatalf("expected 1 consolidated UDP session, got %d", len(result))
+	}
+
+	s := result[0]
+	if s.Protocol != "udp" {
+		t.Errorf("expected protocol=udp, got %q", s.Protocol)
+	}
+	if s.ConnectionCount != 3 {
+		t.Errorf("expected ConnectionCount=3, got %d", s.ConnectionCount)
+	}
+	if s.SourceAddr != "10.0.0.1" {
+		t.Errorf("expected source addr to be IP only, got %q", s.SourceAddr)
+	}
+	if s.BytesTx != 192 {
+		t.Errorf("expected BytesTx=192, got %d", s.BytesTx)
+	}
+	if s.BytesRx != 896 {
+		t.Errorf("expected BytesRx=896, got %d", s.BytesRx)
+	}
+}
+
+func TestConsolidateSessions_MixedGroupsSomeConsolidatedSomeNot(t *testing.T) {
+	now := time.Now()
+	sessions := []*AccessSession{
+		// Group 1: 3 connections to :443 from same IP — should consolidate
+		{
+			SessionID:  "s1",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5000",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now,
+			EndedAt:    now.Add(100 * time.Millisecond),
+		},
+		{
+			SessionID:  "s2",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5001",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now.Add(200 * time.Millisecond),
+			EndedAt:    now.Add(300 * time.Millisecond),
+		},
+		{
+			SessionID:  "s3",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5002",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now.Add(400 * time.Millisecond),
+			EndedAt:    now.Add(500 * time.Millisecond),
+		},
+		// Group 2: 1 connection to :8080 from different IP — should pass through
+		{
+			SessionID:  "s4",
+			ResourceID: 2,
+			SourceAddr: "10.0.0.2:6000",
+			DestAddr:   "192.168.1.200:8080",
+			Protocol:   "tcp",
+			StartedAt:  now.Add(1 * time.Second),
+			EndedAt:    now.Add(2 * time.Second),
+		},
+	}
+
+	result := consolidateSessions(sessions)
+	if len(result) != 2 {
+		t.Fatalf("expected 2 sessions total, got %d", len(result))
+	}
+
+	var consolidated, passthrough *AccessSession
+	for _, s := range result {
+		if s.ConnectionCount > 1 {
+			consolidated = s
+		} else {
+			passthrough = s
+		}
+	}
+
+	if consolidated == nil {
+		t.Fatal("expected a consolidated session")
+	}
+	if consolidated.ConnectionCount != 3 {
+		t.Errorf("consolidated: expected ConnectionCount=3, got %d", consolidated.ConnectionCount)
+	}
+
+	if passthrough == nil {
+		t.Fatal("expected a passthrough session")
+	}
+	if passthrough.SessionID != "s4" {
+		t.Errorf("passthrough: expected session s4, got %s", passthrough.SessionID)
+	}
+}
+
+func TestConsolidateSessions_OverlappingConnections(t *testing.T) {
+	now := time.Now()
+	// Connections that overlap in time (not sequential)
+	sessions := []*AccessSession{
+		{
+			SessionID:  "s1",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5000",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now,
+			EndedAt:    now.Add(5 * time.Second),
+			BytesTx:    100,
+		},
+		{
+			SessionID:  "s2",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5001",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now.Add(1 * time.Second),
+			EndedAt:    now.Add(3 * time.Second),
+			BytesTx:    200,
+		},
+		{
+			SessionID:  "s3",
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:5002",
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now.Add(2 * time.Second),
+			EndedAt:    now.Add(6 * time.Second),
+			BytesTx:    300,
+		},
+	}
+
+	result := consolidateSessions(sessions)
+	if len(result) != 1 {
+		t.Fatalf("expected 1 consolidated session, got %d", len(result))
+	}
+
+	s := result[0]
+	if s.ConnectionCount != 3 {
+		t.Errorf("expected ConnectionCount=3, got %d", s.ConnectionCount)
+	}
+	if s.StartedAt != now {
+		t.Error("expected StartedAt to be earliest")
+	}
+	if s.EndedAt != now.Add(6*time.Second) {
+		t.Error("expected EndedAt to be the latest end time")
+	}
+	if s.BytesTx != 600 {
+		t.Errorf("expected BytesTx=600, got %d", s.BytesTx)
+	}
+}
+
+func TestConsolidateSessions_DoesNotMutateOriginals(t *testing.T) {
+	now := time.Now()
+	s1 := &AccessSession{
+		SessionID:  "s1",
+		ResourceID: 1,
+		SourceAddr: "10.0.0.1:5000",
+		DestAddr:   "192.168.1.100:443",
+		Protocol:   "tcp",
+		StartedAt:  now,
+		EndedAt:    now.Add(100 * time.Millisecond),
+		BytesTx:    100,
+	}
+	s2 := &AccessSession{
+		SessionID:  "s2",
+		ResourceID: 1,
+		SourceAddr: "10.0.0.1:5001",
+		DestAddr:   "192.168.1.100:443",
+		Protocol:   "tcp",
+		StartedAt:  now.Add(200 * time.Millisecond),
+		EndedAt:    now.Add(300 * time.Millisecond),
+		BytesTx:    200,
+	}
+
+	// Save original values
+	origS1Addr := s1.SourceAddr
+	origS1Bytes := s1.BytesTx
+	origS2Addr := s2.SourceAddr
+	origS2Bytes := s2.BytesTx
+
+	_ = consolidateSessions([]*AccessSession{s1, s2})
+
+	if s1.SourceAddr != origS1Addr {
+		t.Errorf("s1.SourceAddr was mutated: %q -> %q", origS1Addr, s1.SourceAddr)
+	}
+	if s1.BytesTx != origS1Bytes {
+		t.Errorf("s1.BytesTx was mutated: %d -> %d", origS1Bytes, s1.BytesTx)
+	}
+	if s2.SourceAddr != origS2Addr {
+		t.Errorf("s2.SourceAddr was mutated: %q -> %q", origS2Addr, s2.SourceAddr)
+	}
+	if s2.BytesTx != origS2Bytes {
+		t.Errorf("s2.BytesTx was mutated: %d -> %d", origS2Bytes, s2.BytesTx)
+	}
+}
+
+func TestConsolidateSessions_ThreeBurstsWithGaps(t *testing.T) {
+	now := time.Now()
+
+	sessions := make([]*AccessSession, 0, 9)
+
+	// Burst 1: 3 connections at t=0
+	for i := 0; i < 3; i++ {
+		sessions = append(sessions, &AccessSession{
+			SessionID:  generateSessionID(),
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:" + string(rune('A'+i)),
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now.Add(time.Duration(i*100) * time.Millisecond),
+			EndedAt:    now.Add(time.Duration(i*100+50) * time.Millisecond),
+		})
+	}
+
+	// Burst 2: 3 connections at t=20s (well past the 5s gap)
+	for i := 0; i < 3; i++ {
+		sessions = append(sessions, &AccessSession{
+			SessionID:  generateSessionID(),
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:" + string(rune('D'+i)),
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now.Add(20*time.Second + time.Duration(i*100)*time.Millisecond),
+			EndedAt:    now.Add(20*time.Second + time.Duration(i*100+50)*time.Millisecond),
+		})
+	}
+
+	// Burst 3: 3 connections at t=40s
+	for i := 0; i < 3; i++ {
+		sessions = append(sessions, &AccessSession{
+			SessionID:  generateSessionID(),
+			ResourceID: 1,
+			SourceAddr: "10.0.0.1:" + string(rune('G'+i)),
+			DestAddr:   "192.168.1.100:443",
+			Protocol:   "tcp",
+			StartedAt:  now.Add(40*time.Second + time.Duration(i*100)*time.Millisecond),
+			EndedAt:    now.Add(40*time.Second + time.Duration(i*100+50)*time.Millisecond),
+		})
+	}
+
+	result := consolidateSessions(sessions)
+	if len(result) != 3 {
+		t.Fatalf("expected 3 consolidated sessions (3 bursts), got %d", len(result))
+	}
+
+	for _, s := range result {
+		if s.ConnectionCount != 3 {
+			t.Errorf("expected each burst to have ConnectionCount=3, got %d (started=%v)", s.ConnectionCount, s.StartedAt)
+		}
+	}
+}
+
+func TestFinalizeMergedSourceAddr(t *testing.T) {
+	s := &AccessSession{SourceAddr: "10.0.0.1:5000"}
+	ports := map[string]struct{}{"10.0.0.1:5000": {}}
+	finalizeMergedSourceAddr(s, "10.0.0.1", ports)
+	if s.SourceAddr != "10.0.0.1:5000" {
+		t.Errorf("single port: expected addr preserved, got %q", s.SourceAddr)
+	}
+
+	s2 := &AccessSession{SourceAddr: "10.0.0.1:5000"}
+	ports2 := map[string]struct{}{"10.0.0.1:5000": {}, "10.0.0.1:5001": {}}
+	finalizeMergedSourceAddr(s2, "10.0.0.1", ports2)
+	if s2.SourceAddr != "10.0.0.1" {
+		t.Errorf("multiple ports: expected IP only, got %q", s2.SourceAddr)
+	}
+}
+
+func TestCloneSession(t *testing.T) {
+	original := &AccessSession{
+		SessionID:  "test",
+		ResourceID: 42,
+		SourceAddr: "1.2.3.4:100",
+		DestAddr:   "5.6.7.8:443",
+		Protocol:   "tcp",
+		BytesTx:    999,
+	}
+
+	clone := cloneSession(original)
+
+	if clone == original {
+		t.Error("clone should be a different pointer")
+	}
+	if clone.SessionID != original.SessionID {
+		t.Error("clone should have same SessionID")
+	}
+
+	// Mutating clone should not affect original
+	clone.BytesTx = 0
+	clone.SourceAddr = "changed"
+	if original.BytesTx != 999 {
+		t.Error("mutating clone affected original BytesTx")
+	}
+	if original.SourceAddr != "1.2.3.4:100" {
+		t.Error("mutating clone affected original SourceAddr")
+	}
+}

netstack2/handlers.go

@@ -0,0 +1,776 @@
+package netstack2
+
+import (
+	"context"
+	"fmt"
+	"io"
+	"net"
+	"net/netip"
+	"os/exec"
+	"sync"
+	"time"
+
+	"github.com/fosrl/newt/logger"
+	"golang.org/x/net/icmp"
+	"golang.org/x/net/ipv4"
+	"gvisor.dev/gvisor/pkg/tcpip"
+	"gvisor.dev/gvisor/pkg/tcpip/adapters/gonet"
+	"gvisor.dev/gvisor/pkg/tcpip/checksum"
+	"gvisor.dev/gvisor/pkg/tcpip/header"
+	"gvisor.dev/gvisor/pkg/tcpip/stack"
+	"gvisor.dev/gvisor/pkg/tcpip/transport/tcp"
+	"gvisor.dev/gvisor/pkg/tcpip/transport/udp"
+	"gvisor.dev/gvisor/pkg/waiter"
+)
+
+const (
+	// defaultWndSize if set to zero, the default
+	// receive window buffer size is used instead.
+	defaultWndSize = 0
+
+	// maxConnAttempts specifies the maximum number
+	// of in-flight tcp connection attempts.
+	maxConnAttempts = 2 << 10
+
+	// tcpKeepaliveCount is the maximum number of
+	// TCP keep-alive probes to send before giving up
+	// and killing the connection if no response is
+	// obtained from the other end.
+	tcpKeepaliveCount = 9
+
+	// tcpKeepaliveIdle specifies the time a connection
+	// must remain idle before the first TCP keepalive
+	// packet is sent. Once this time is reached,
+	// tcpKeepaliveInterval option is used instead.
+	tcpKeepaliveIdle = 60 * time.Second
+
+	// tcpKeepaliveInterval specifies the interval
+	// time between sending TCP keepalive packets.
+	tcpKeepaliveInterval = 30 * time.Second
+
+	// tcpConnectTimeout is the default timeout for TCP handshakes.
+	tcpConnectTimeout = 5 * time.Second
+
+	// tcpWaitTimeout implements a TCP half-close timeout.
+	tcpWaitTimeout = 60 * time.Second
+
+	// udpSessionTimeout is the default timeout for UDP sessions.
+	udpSessionTimeout = 60 * time.Second
+
+	// Buffer size for copying data
+	bufferSize = 32 * 1024
+
+	// icmpTimeout is the default timeout for ICMP ping requests.
+	icmpTimeout = 5 * time.Second
+)
+
+// TCPHandler handles TCP connections from netstack
+type TCPHandler struct {
+	stack        *stack.Stack
+	proxyHandler *ProxyHandler
+}
+
+// UDPHandler handles UDP connections from netstack
+type UDPHandler struct {
+	stack        *stack.Stack
+	proxyHandler *ProxyHandler
+}
+
+// ICMPHandler handles ICMP packets from netstack
+type ICMPHandler struct {
+	stack        *stack.Stack
+	proxyHandler *ProxyHandler
+}
+
+// NewTCPHandler creates a new TCP handler
+func NewTCPHandler(s *stack.Stack, ph *ProxyHandler) *TCPHandler {
+	return &TCPHandler{stack: s, proxyHandler: ph}
+}
+
+// NewUDPHandler creates a new UDP handler
+func NewUDPHandler(s *stack.Stack, ph *ProxyHandler) *UDPHandler {
+	return &UDPHandler{stack: s, proxyHandler: ph}
+}
+
+// NewICMPHandler creates a new ICMP handler
+func NewICMPHandler(s *stack.Stack, ph *ProxyHandler) *ICMPHandler {
+	return &ICMPHandler{stack: s, proxyHandler: ph}
+}
+
+// InstallTCPHandler installs the TCP forwarder on the stack
+func (h *TCPHandler) InstallTCPHandler() error {
+	tcpForwarder := tcp.NewForwarder(h.stack, defaultWndSize, maxConnAttempts, func(r *tcp.ForwarderRequest) {
+		var (
+			wq  waiter.Queue
+			ep  tcpip.Endpoint
+			err tcpip.Error
+			id  = r.ID()
+		)
+
+		// Perform a TCP three-way handshake
+		ep, err = r.CreateEndpoint(&wq)
+		if err != nil {
+			// RST: prevent potential half-open TCP connection leak
+			r.Complete(true)
+			return
+		}
+		defer r.Complete(false)
+
+		// Set socket options
+		setTCPSocketOptions(h.stack, ep)
+
+		// Create TCP connection from netstack endpoint
+		netstackConn := gonet.NewTCPConn(&wq, ep)
+
+		// Handle the connection in a goroutine
+		go h.handleTCPConn(netstackConn, id)
+	})
+
+	h.stack.SetTransportProtocolHandler(tcp.ProtocolNumber, tcpForwarder.HandlePacket)
+	return nil
+}
+
+// handleTCPConn handles a TCP connection by proxying it to the actual target
+func (h *TCPHandler) handleTCPConn(netstackConn *gonet.TCPConn, id stack.TransportEndpointID) {
+	// Extract source and target address from the connection ID first so they
+	// are available for HTTP routing before any defer is set up.
+	srcIP := id.RemoteAddress.String()
+	srcPort := id.RemotePort
+	dstIP := id.LocalAddress.String()
+	dstPort := id.LocalPort
+
+	// Drop connection if blocking is enabled
+	if h.proxyHandler != nil && h.proxyHandler.IsBlocked() {
+		logger.Debug("TCP Forwarder: connection blocked: %s:%d -> %s:%d", srcIP, srcPort, dstIP, dstPort)
+		netstackConn.Close()
+		return
+	}
+
+	// For HTTP/HTTPS ports, look up the matching subnet rule. If the rule has
+	// Protocol configured, hand the connection off to the HTTP handler which
+	// takes full ownership of the lifecycle (the defer close must not be
+	// installed before this point).
+	if (dstPort == 80 || dstPort == 443) && h.proxyHandler != nil && h.proxyHandler.httpHandler != nil {
+		srcAddr, _ := netip.ParseAddr(srcIP)
+		dstAddr, _ := netip.ParseAddr(dstIP)
+		rule := h.proxyHandler.subnetLookup.Match(srcAddr, dstAddr, dstPort, tcp.ProtocolNumber)
+		if rule != nil && rule.Protocol != "" && len(rule.HTTPTargets) > 0 {
+			logger.Info("TCP Forwarder: Routing %s:%d -> %s:%d to HTTP handler (%s)",
+				srcIP, srcPort, dstIP, dstPort, rule.Protocol)
+			h.proxyHandler.httpHandler.HandleConn(netstackConn, rule)
+			return
+		}
+		// Otherwise fall through to raw TCP forwarding (e.g. CIDR resources
+		// that happen to use port 80/443 without HTTP configuration).
+	}
+
+	defer netstackConn.Close()
+
+	logger.Info("TCP Forwarder: Handling connection %s:%d -> %s:%d", srcIP, srcPort, dstIP, dstPort)
+
+	// Check if there's a destination rewrite for this connection (e.g., localhost targets)
+	actualDstIP := dstIP
+	if h.proxyHandler != nil {
+		if rewrittenAddr, ok := h.proxyHandler.LookupDestinationRewrite(srcIP, dstIP, dstPort, uint8(tcp.ProtocolNumber)); ok {
+			actualDstIP = rewrittenAddr.String()
+			logger.Info("TCP Forwarder: Using rewritten destination %s (original: %s)", actualDstIP, dstIP)
+		}
+	}
+
+	targetAddr := fmt.Sprintf("%s:%d", actualDstIP, dstPort)
+
+	// Look up resource ID and start access session if applicable
+	var accessSessionID string
+	if h.proxyHandler != nil {
+		resourceId := h.proxyHandler.LookupResourceId(srcIP, dstIP, dstPort, uint8(tcp.ProtocolNumber))
+		if resourceId != 0 {
+			if al := h.proxyHandler.GetAccessLogger(); al != nil {
+				srcAddr := fmt.Sprintf("%s:%d", srcIP, srcPort)
+				accessSessionID = al.StartTCPSession(resourceId, srcAddr, targetAddr)
+			}
+		}
+	}
+
+	// Create context with timeout for connection establishment
+	ctx, cancel := context.WithTimeout(context.Background(), tcpConnectTimeout)
+	defer cancel()
+
+	// Dial the actual target using standard net package
+	var d net.Dialer
+	targetConn, err := d.DialContext(ctx, "tcp", targetAddr)
+	if err != nil {
+		logger.Info("TCP Forwarder: Failed to connect to %s: %v", targetAddr, err)
+		// End access session on connection failure
+		if accessSessionID != "" {
+			if al := h.proxyHandler.GetAccessLogger(); al != nil {
+				al.EndTCPSession(accessSessionID)
+			}
+		}
+		// Connection failed, netstack will handle RST
+		return
+	}
+	defer targetConn.Close()
+
+	// End access session when connection closes
+	if accessSessionID != "" {
+		defer func() {
+			if al := h.proxyHandler.GetAccessLogger(); al != nil {
+				al.EndTCPSession(accessSessionID)
+			}
+		}()
+	}
+
+	logger.Info("TCP Forwarder: Successfully connected to %s, starting bidirectional copy", targetAddr)
+
+	// Bidirectional copy between netstack and target
+	pipeTCP(netstackConn, targetConn)
+}
+
+// pipeTCP copies data bidirectionally between two connections
+func pipeTCP(origin, remote net.Conn) {
+	wg := sync.WaitGroup{}
+	wg.Add(2)
+
+	go unidirectionalStreamTCP(remote, origin, "origin->remote", &wg)
+	go unidirectionalStreamTCP(origin, remote, "remote->origin", &wg)
+
+	wg.Wait()
+}
+
+// unidirectionalStreamTCP copies data in one direction
+func unidirectionalStreamTCP(dst, src net.Conn, dir string, wg *sync.WaitGroup) {
+	defer wg.Done()
+
+	buf := make([]byte, bufferSize)
+	_, _ = io.CopyBuffer(dst, src, buf)
+
+	// Do the upload/download side TCP half-close
+	if cr, ok := src.(interface{ CloseRead() error }); ok {
+		cr.CloseRead()
+	}
+	if cw, ok := dst.(interface{ CloseWrite() error }); ok {
+		cw.CloseWrite()
+	}
+
+	// Set TCP half-close timeout
+	dst.SetReadDeadline(time.Now().Add(tcpWaitTimeout))
+}
+
+// setTCPSocketOptions sets TCP socket options for better performance
+func setTCPSocketOptions(s *stack.Stack, ep tcpip.Endpoint) {
+	// TCP keepalive options
+	ep.SocketOptions().SetKeepAlive(true)
+
+	idle := tcpip.KeepaliveIdleOption(tcpKeepaliveIdle)
+	ep.SetSockOpt(&idle)
+
+	interval := tcpip.KeepaliveIntervalOption(tcpKeepaliveInterval)
+	ep.SetSockOpt(&interval)
+
+	ep.SetSockOptInt(tcpip.KeepaliveCountOption, tcpKeepaliveCount)
+
+	// TCP send/recv buffer size
+	var ss tcpip.TCPSendBufferSizeRangeOption
+	if err := s.TransportProtocolOption(tcp.ProtocolNumber, &ss); err == nil {
+		ep.SocketOptions().SetSendBufferSize(int64(ss.Default), false)
+	}
+
+	var rs tcpip.TCPReceiveBufferSizeRangeOption
+	if err := s.TransportProtocolOption(tcp.ProtocolNumber, &rs); err == nil {
+		ep.SocketOptions().SetReceiveBufferSize(int64(rs.Default), false)
+	}
+}
+
+// InstallUDPHandler installs the UDP forwarder on the stack
+func (h *UDPHandler) InstallUDPHandler() error {
+	udpForwarder := udp.NewForwarder(h.stack, func(r *udp.ForwarderRequest) {
+		var (
+			wq waiter.Queue
+			id = r.ID()
+		)
+
+		ep, err := r.CreateEndpoint(&wq)
+		if err != nil {
+			return
+		}
+
+		// Create UDP connection from netstack endpoint
+		netstackConn := gonet.NewUDPConn(&wq, ep)
+
+		// Handle the connection in a goroutine
+		go h.handleUDPConn(netstackConn, id)
+	})
+
+	h.stack.SetTransportProtocolHandler(udp.ProtocolNumber, udpForwarder.HandlePacket)
+	return nil
+}
+
+// handleUDPConn handles a UDP connection by proxying it to the actual target
+func (h *UDPHandler) handleUDPConn(netstackConn *gonet.UDPConn, id stack.TransportEndpointID) {
+	defer netstackConn.Close()
+
+	// Extract source and target address from the connection ID
+	srcIP := id.RemoteAddress.String()
+	srcPort := id.RemotePort
+	dstIP := id.LocalAddress.String()
+	dstPort := id.LocalPort
+
+	logger.Info("UDP Forwarder: Handling connection %s:%d -> %s:%d", srcIP, srcPort, dstIP, dstPort)
+
+	// Drop connection if blocking is enabled
+	if h.proxyHandler != nil && h.proxyHandler.IsBlocked() {
+		logger.Debug("UDP Forwarder: connection blocked: %s:%d -> %s:%d", srcIP, srcPort, dstIP, dstPort)
+		return
+	}
+
+	// Check if there's a destination rewrite for this connection (e.g., localhost targets)
+	actualDstIP := dstIP
+	if h.proxyHandler != nil {
+		if rewrittenAddr, ok := h.proxyHandler.LookupDestinationRewrite(srcIP, dstIP, dstPort, uint8(udp.ProtocolNumber)); ok {
+			actualDstIP = rewrittenAddr.String()
+			logger.Info("UDP Forwarder: Using rewritten destination %s (original: %s)", actualDstIP, dstIP)
+		}
+	}
+
+	targetAddr := fmt.Sprintf("%s:%d", actualDstIP, dstPort)
+
+	// Look up resource ID and start access session if applicable
+	var accessSessionID string
+	if h.proxyHandler != nil {
+		resourceId := h.proxyHandler.LookupResourceId(srcIP, dstIP, dstPort, uint8(udp.ProtocolNumber))
+		if resourceId != 0 {
+			if al := h.proxyHandler.GetAccessLogger(); al != nil {
+				srcAddr := fmt.Sprintf("%s:%d", srcIP, srcPort)
+				accessSessionID = al.TrackUDPSession(resourceId, srcAddr, targetAddr)
+			}
+		}
+	}
+
+	// End access session when UDP handler returns (timeout or error)
+	if accessSessionID != "" {
+		defer func() {
+			if al := h.proxyHandler.GetAccessLogger(); al != nil {
+				al.EndUDPSession(accessSessionID)
+			}
+		}()
+	}
+
+	// Resolve target address
+	remoteUDPAddr, err := net.ResolveUDPAddr("udp", targetAddr)
+	if err != nil {
+		logger.Info("UDP Forwarder: Failed to resolve %s: %v", targetAddr, err)
+		return
+	}
+
+	// Resolve client address (for sending responses back)
+	clientAddr, err := net.ResolveUDPAddr("udp", fmt.Sprintf("%s:%d", srcIP, srcPort))
+	if err != nil {
+		logger.Info("UDP Forwarder: Failed to resolve client %s:%d: %v", srcIP, srcPort, err)
+		return
+	}
+
+	// Create unconnected UDP socket (so we can use WriteTo)
+	targetConn, err := net.ListenUDP("udp", nil)
+	if err != nil {
+		logger.Info("UDP Forwarder: Failed to create UDP socket: %v", err)
+		return
+	}
+	defer targetConn.Close()
+
+	logger.Info("UDP Forwarder: Successfully created UDP socket for %s, starting bidirectional copy", targetAddr)
+
+	// Bidirectional copy between netstack and target
+	pipeUDP(netstackConn, targetConn, remoteUDPAddr, clientAddr, udpSessionTimeout)
+}
+
+// pipeUDP copies UDP packets bidirectionally
+func pipeUDP(origin, remote net.PacketConn, serverAddr, clientAddr net.Addr, timeout time.Duration) {
+	wg := sync.WaitGroup{}
+	wg.Add(2)
+
+	// Read from origin (netstack), write to remote (target server)
+	go unidirectionalPacketStream(remote, origin, serverAddr, "origin->remote", &wg, timeout)
+	// Read from remote (target server), write to origin (netstack) with client address
+	go unidirectionalPacketStream(origin, remote, clientAddr, "remote->origin", &wg, timeout)
+
+	wg.Wait()
+}
+
+// unidirectionalPacketStream copies packets in one direction
+func unidirectionalPacketStream(dst, src net.PacketConn, to net.Addr, dir string, wg *sync.WaitGroup, timeout time.Duration) {
+	defer wg.Done()
+
+	logger.Info("UDP %s: Starting packet stream (to=%v)", dir, to)
+	err := copyPacketData(dst, src, to, timeout)
+	if err != nil {
+		logger.Info("UDP %s: Stream ended with error: %v", dir, err)
+	} else {
+		logger.Info("UDP %s: Stream ended (timeout)", dir)
+	}
+}
+
+// copyPacketData copies UDP packet data with timeout
+func copyPacketData(dst, src net.PacketConn, to net.Addr, timeout time.Duration) error {
+	buf := make([]byte, 65535) // Max UDP packet size
+
+	for {
+		src.SetReadDeadline(time.Now().Add(timeout))
+		n, srcAddr, err := src.ReadFrom(buf)
+		if ne, ok := err.(net.Error); ok && ne.Timeout() {
+			return nil // ignore I/O timeout
+		} else if err == io.EOF {
+			return nil // ignore EOF
+		} else if err != nil {
+			return err
+		}
+
+		logger.Info("UDP copyPacketData: Read %d bytes from %v", n, srcAddr)
+
+		// Determine write destination
+		writeAddr := to
+		if writeAddr == nil {
+			// If no destination specified, use the source address from the packet
+			writeAddr = srcAddr
+		}
+
+		written, err := dst.WriteTo(buf[:n], writeAddr)
+		if err != nil {
+			logger.Info("UDP copyPacketData: Write error to %v: %v", writeAddr, err)
+			return err
+		}
+		logger.Info("UDP copyPacketData: Wrote %d bytes to %v", written, writeAddr)
+
+		dst.SetReadDeadline(time.Now().Add(timeout))
+	}
+}
+
+// InstallICMPHandler installs the ICMP handler on the stack
+func (h *ICMPHandler) InstallICMPHandler() error {
+	h.stack.SetTransportProtocolHandler(header.ICMPv4ProtocolNumber, h.handleICMPPacket)
+	logger.Debug("ICMP Handler: Installed ICMP protocol handler")
+	return nil
+}
+
+// handleICMPPacket handles incoming ICMP packets
+func (h *ICMPHandler) handleICMPPacket(id stack.TransportEndpointID, pkt *stack.PacketBuffer) bool {
+	logger.Debug("ICMP Handler: Received ICMP packet from %s to %s", id.RemoteAddress, id.LocalAddress)
+
+	// Get the ICMP header from the packet
+	icmpData := pkt.TransportHeader().Slice()
+	if len(icmpData) < header.ICMPv4MinimumSize {
+		logger.Debug("ICMP Handler: Packet too small for ICMP header: %d bytes", len(icmpData))
+		return false
+	}
+
+	icmpHdr := header.ICMPv4(icmpData)
+	icmpType := icmpHdr.Type()
+	icmpCode := icmpHdr.Code()
+
+	logger.Debug("ICMP Handler: Type=%d, Code=%d, Ident=%d, Seq=%d",
+		icmpType, icmpCode, icmpHdr.Ident(), icmpHdr.Sequence())
+
+	// Only handle Echo Request (ping)
+	if icmpType != header.ICMPv4Echo {
+		logger.Debug("ICMP Handler: Ignoring non-echo ICMP type: %d", icmpType)
+		return false
+	}
+
+	// Extract source and destination addresses
+	srcIP := id.RemoteAddress.String()
+	dstIP := id.LocalAddress.String()
+
+	logger.Info("ICMP Handler: Echo Request from %s to %s (ident=%d, seq=%d)",
+		srcIP, dstIP, icmpHdr.Ident(), icmpHdr.Sequence())
+
+	// Convert to netip.Addr for subnet matching
+	srcAddr, err := netip.ParseAddr(srcIP)
+	if err != nil {
+		logger.Debug("ICMP Handler: Failed to parse source IP %s: %v", srcIP, err)
+		return false
+	}
+	dstAddr, err := netip.ParseAddr(dstIP)
+	if err != nil {
+		logger.Debug("ICMP Handler: Failed to parse dest IP %s: %v", dstIP, err)
+		return false
+	}
+
+	// Check subnet rules (use port 0 for ICMP since it doesn't have ports)
+	if h.proxyHandler == nil {
+		logger.Debug("ICMP Handler: No proxy handler configured")
+		return false
+	}
+
+	matchedRule := h.proxyHandler.subnetLookup.Match(srcAddr, dstAddr, 0, header.ICMPv4ProtocolNumber)
+	if matchedRule == nil {
+		logger.Debug("ICMP Handler: No matching subnet rule for %s -> %s", srcIP, dstIP)
+		return false
+	}
+
+	logger.Info("ICMP Handler: Matched subnet rule for %s -> %s", srcIP, dstIP)
+
+	// Determine actual destination (with possible rewrite)
+	actualDstIP := dstIP
+	if matchedRule.RewriteTo != "" {
+		resolvedAddr, err := h.proxyHandler.resolveRewriteAddress(matchedRule.RewriteTo)
+		if err != nil {
+			logger.Info("ICMP Handler: Failed to resolve rewrite address %s: %v", matchedRule.RewriteTo, err)
+		} else {
+			actualDstIP = resolvedAddr.String()
+			logger.Info("ICMP Handler: Using rewritten destination %s (original: %s)", actualDstIP, dstIP)
+		}
+	}
+
+	// Get the full ICMP payload (including the data after the header)
+	icmpPayload := pkt.Data().AsRange().ToSlice()
+
+	// Handle the ping in a goroutine to avoid blocking
+	go h.proxyPing(srcIP, dstIP, actualDstIP, icmpHdr.Ident(), icmpHdr.Sequence(), icmpPayload)
+
+	return true
+}
+
+// proxyPing sends a ping to the actual destination and injects the reply back
+func (h *ICMPHandler) proxyPing(srcIP, originalDstIP, actualDstIP string, ident, seq uint16, payload []byte) {
+	logger.Debug("ICMP Handler: Proxying ping from %s to %s (actual: %s), ident=%d, seq=%d",
+		srcIP, originalDstIP, actualDstIP, ident, seq)
+
+	// Try three methods in order: ip4:icmp -> udp4 -> ping command
+	// Track which method succeeded so we can handle identifier matching correctly
+	method, success := h.tryICMPMethods(actualDstIP, ident, seq, payload)
+
+	if !success {
+		logger.Info("ICMP Handler: All ping methods failed for %s", actualDstIP)
+		return
+	}
+
+	logger.Info("ICMP Handler: Ping successful to %s using %s, injecting reply (ident=%d, seq=%d)",
+		actualDstIP, method, ident, seq)
+
+	// Build the reply packet to inject back into the netstack
+	// The reply should appear to come from the original destination (before rewrite)
+	h.injectICMPReply(srcIP, originalDstIP, ident, seq, payload)
+}
+
+// tryICMPMethods tries all available ICMP methods in order
+func (h *ICMPHandler) tryICMPMethods(actualDstIP string, ident, seq uint16, payload []byte) (string, bool) {
+	if h.tryRawICMP(actualDstIP, ident, seq, payload, false) {
+		return "raw ICMP", true
+	}
+	if h.tryUnprivilegedICMP(actualDstIP, ident, seq, payload) {
+		return "unprivileged ICMP", true
+	}
+	if h.tryPingCommand(actualDstIP, ident, seq, payload) {
+		return "ping command", true
+	}
+	return "", false
+}
+
+// tryRawICMP attempts to ping using raw ICMP sockets (requires CAP_NET_RAW or root)
+func (h *ICMPHandler) tryRawICMP(actualDstIP string, ident, seq uint16, payload []byte, ignoreIdent bool) bool {
+	conn, err := icmp.ListenPacket("ip4:icmp", "0.0.0.0")
+	if err != nil {
+		logger.Debug("ICMP Handler: Raw ICMP socket not available: %v", err)
+		return false
+	}
+	defer conn.Close()
+
+	logger.Debug("ICMP Handler: Using raw ICMP socket")
+	return h.sendAndReceiveICMP(conn, actualDstIP, ident, seq, payload, false, ignoreIdent)
+}
+
+// tryUnprivilegedICMP attempts to ping using unprivileged ICMP (requires ping_group_range configured)
+func (h *ICMPHandler) tryUnprivilegedICMP(actualDstIP string, ident, seq uint16, payload []byte) bool {
+	conn, err := icmp.ListenPacket("udp4", "0.0.0.0")
+	if err != nil {
+		logger.Debug("ICMP Handler: Unprivileged ICMP socket not available: %v", err)
+		return false
+	}
+	defer conn.Close()
+
+	logger.Debug("ICMP Handler: Using unprivileged ICMP socket")
+	// Unprivileged ICMP doesn't let us control the identifier, so we ignore it in matching
+	return h.sendAndReceiveICMP(conn, actualDstIP, ident, seq, payload, true, true)
+}
+
+// sendAndReceiveICMP sends an ICMP echo request and waits for the reply
+func (h *ICMPHandler) sendAndReceiveICMP(conn *icmp.PacketConn, actualDstIP string, ident, seq uint16, payload []byte, isUnprivileged bool, ignoreIdent bool) bool {
+	// Build the ICMP echo request message
+	echoMsg := &icmp.Message{
+		Type: ipv4.ICMPTypeEcho,
+		Code: 0,
+		Body: &icmp.Echo{
+			ID:   int(ident),
+			Seq:  int(seq),
+			Data: payload,
+		},
+	}
+
+	msgBytes, err := echoMsg.Marshal(nil)
+	if err != nil {
+		logger.Debug("ICMP Handler: Failed to marshal ICMP message: %v", err)
+		return false
+	}
+
+	// Resolve destination address based on socket type
+	var writeErr error
+	if isUnprivileged {
+		// For unprivileged ICMP, use UDP-style addressing
+		udpAddr := &net.UDPAddr{IP: net.ParseIP(actualDstIP)}
+		logger.Debug("ICMP Handler: Sending ping to %s (unprivileged)", udpAddr.String())
+		conn.SetDeadline(time.Now().Add(icmpTimeout))
+		_, writeErr = conn.WriteTo(msgBytes, udpAddr)
+	} else {
+		// For raw ICMP, use IP addressing
+		dst, err := net.ResolveIPAddr("ip4", actualDstIP)
+		if err != nil {
+			logger.Debug("ICMP Handler: Failed to resolve destination %s: %v", actualDstIP, err)
+			return false
+		}
+		logger.Debug("ICMP Handler: Sending ping to %s (raw)", dst.String())
+		conn.SetDeadline(time.Now().Add(icmpTimeout))
+		_, writeErr = conn.WriteTo(msgBytes, dst)
+	}
+
+	if writeErr != nil {
+		logger.Debug("ICMP Handler: Failed to send ping to %s: %v", actualDstIP, writeErr)
+		return false
+	}
+
+	logger.Debug("ICMP Handler: Ping sent to %s, waiting for reply (ident=%d, seq=%d)", actualDstIP, ident, seq)
+
+	// Wait for reply - loop to filter out non-matching packets
+	replyBuf := make([]byte, 1500)
+
+	for {
+		n, peer, err := conn.ReadFrom(replyBuf)
+		if err != nil {
+			logger.Debug("ICMP Handler: Failed to receive ping reply from %s: %v", actualDstIP, err)
+			return false
+		}
+
+		logger.Debug("ICMP Handler: Received %d bytes from %s", n, peer.String())
+
+		// Parse the reply
+		replyMsg, err := icmp.ParseMessage(1, replyBuf[:n])
+		if err != nil {
+			logger.Debug("ICMP Handler: Failed to parse ICMP message: %v", err)
+			continue
+		}
+
+		// Check if it's an echo reply (type 0), not an echo request (type 8)
+		if replyMsg.Type != ipv4.ICMPTypeEchoReply {
+			logger.Debug("ICMP Handler: Received non-echo-reply type: %v, continuing to wait", replyMsg.Type)
+			continue
+		}
+
+		reply, ok := replyMsg.Body.(*icmp.Echo)
+		if !ok {
+			logger.Debug("ICMP Handler: Invalid echo reply body type, continuing to wait")
+			continue
+		}
+
+		// Verify the sequence matches what we sent
+		// For unprivileged ICMP, the kernel controls the identifier, so we only check sequence
+		if reply.Seq != int(seq) {
+			logger.Debug("ICMP Handler: Reply seq mismatch: got seq=%d, want seq=%d", reply.Seq, seq)
+			continue
+		}
+
+		if !ignoreIdent && reply.ID != int(ident) {
+			logger.Debug("ICMP Handler: Reply ident mismatch: got ident=%d, want ident=%d", reply.ID, ident)
+			continue
+		}
+
+		// Found matching reply
+		logger.Debug("ICMP Handler: Received valid echo reply")
+		return true
+	}
+}
+
+// tryPingCommand attempts to ping using the system ping command (always works, but less control)
+func (h *ICMPHandler) tryPingCommand(actualDstIP string, ident, seq uint16, payload []byte) bool {
+	logger.Debug("ICMP Handler: Attempting to use system ping command")
+
+	ctx, cancel := context.WithTimeout(context.Background(), icmpTimeout)
+	defer cancel()
+
+	// Send one ping with timeout
+	// -c 1: count = 1 packet
+	// -W 5: timeout = 5 seconds
+	// -q: quiet output (just summary)
+	cmd := exec.CommandContext(ctx, "ping", "-c", "1", "-W", "5", "-q", actualDstIP)
+	output, err := cmd.CombinedOutput()
+
+	if err != nil {
+		logger.Debug("ICMP Handler: System ping command failed: %v, output: %s", err, string(output))
+		return false
+	}
+
+	logger.Debug("ICMP Handler: System ping command succeeded")
+	return true
+}
+
+// injectICMPReply creates an ICMP echo reply packet and queues it to be sent back through the tunnel
+func (h *ICMPHandler) injectICMPReply(dstIP, srcIP string, ident, seq uint16, payload []byte) {
+	logger.Debug("ICMP Handler: Creating reply from %s to %s (ident=%d, seq=%d)",
+		srcIP, dstIP, ident, seq)
+
+	// Parse addresses
+	srcAddr, err := netip.ParseAddr(srcIP)
+	if err != nil {
+		logger.Info("ICMP Handler: Failed to parse source IP for reply: %v", err)
+		return
+	}
+	dstAddr, err := netip.ParseAddr(dstIP)
+	if err != nil {
+		logger.Info("ICMP Handler: Failed to parse dest IP for reply: %v", err)
+		return
+	}
+
+	// Calculate total packet size
+	ipHeaderLen := header.IPv4MinimumSize
+	icmpHeaderLen := header.ICMPv4MinimumSize
+	totalLen := ipHeaderLen + icmpHeaderLen + len(payload)
+
+	// Create the packet buffer
+	pkt := make([]byte, totalLen)
+
+	// Build IPv4 header
+	ipHdr := header.IPv4(pkt[:ipHeaderLen])
+	ipHdr.Encode(&header.IPv4Fields{
+		TotalLength: uint16(totalLen),
+		TTL:         64,
+		Protocol:    uint8(header.ICMPv4ProtocolNumber),
+		SrcAddr:     tcpip.AddrFrom4(srcAddr.As4()),
+		DstAddr:     tcpip.AddrFrom4(dstAddr.As4()),
+	})
+	ipHdr.SetChecksum(^ipHdr.CalculateChecksum())
+
+	// Build ICMP header
+	icmpHdr := header.ICMPv4(pkt[ipHeaderLen : ipHeaderLen+icmpHeaderLen])
+	icmpHdr.SetType(header.ICMPv4EchoReply)
+	icmpHdr.SetCode(0)
+	icmpHdr.SetIdent(ident)
+	icmpHdr.SetSequence(seq)
+
+	// Copy payload
+	copy(pkt[ipHeaderLen+icmpHeaderLen:], payload)
+
+	// Calculate ICMP checksum (covers ICMP header + payload)
+	icmpHdr.SetChecksum(0)
+	icmpData := pkt[ipHeaderLen:]
+	icmpHdr.SetChecksum(^checksum.Checksum(icmpData, 0))
+
+	logger.Debug("ICMP Handler: Built reply packet, total length=%d", totalLen)
+
+	// Queue the packet to be sent back through the tunnel
+	if h.proxyHandler != nil {
+		if h.proxyHandler.QueueICMPReply(pkt) {
+			logger.Info("ICMP Handler: Queued echo reply packet for transmission")
+		} else {
+			logger.Info("ICMP Handler: Failed to queue echo reply packet")
+		}
+	} else {
+		logger.Info("ICMP Handler: Cannot queue reply - proxy handler not available")
+	}
+}

netstack2/http_handler.go

@@ -0,0 +1,420 @@
+package netstack2
+
+import (
+	"bufio"
+	"context"
+	"crypto/tls"
+	"errors"
+	"fmt"
+	"net"
+	"net/http"
+	"net/http/httputil"
+	"net/url"
+	"sync"
+	"time"
+
+	"github.com/fosrl/newt/logger"
+	"gvisor.dev/gvisor/pkg/tcpip/stack"
+)
+
+// ---------------------------------------------------------------------------
+// HTTPTarget
+// ---------------------------------------------------------------------------
+
+// HTTPTarget describes a single downstream HTTP or HTTPS service that the
+// proxy should forward requests to.
+type HTTPTarget struct {
+	DestAddr string `json:"destAddr"` // IP address or hostname of the downstream service
+	DestPort uint16 `json:"destPort"` // TCP port of the downstream service
+	Scheme   string `json:"scheme"`   // When true the outbound leg uses HTTPS
+}
+
+// ---------------------------------------------------------------------------
+// HTTPHandler
+// ---------------------------------------------------------------------------
+
+// HTTPHandler intercepts TCP connections from the netstack forwarder on ports
+// 80 and 443 and services them as HTTP or HTTPS, reverse-proxying each request
+// to downstream targets specified by the matching SubnetRule.
+//
+// HTTP and raw TCP are fully separate: a connection is only routed here when
+// its SubnetRule has Protocol set ("http" or "https"). All other connections
+// on those ports fall through to the normal raw-TCP path.
+//
+// Incoming TLS termination (Protocol == "https") is performed per-connection
+// using the certificate and key stored in the rule, so different subnet rules
+// can present different certificates without sharing any state.
+//
+// Outbound connections to downstream targets honour HTTPTarget.UseHTTPS
+// independently of the incoming protocol.
+type HTTPHandler struct {
+	stack         *stack.Stack
+	proxyHandler  *ProxyHandler
+	requestLogger *HTTPRequestLogger
+
+	listener *chanListener
+	server   *http.Server
+
+	// proxyCache holds pre-built *httputil.ReverseProxy values keyed by the
+	// canonical target URL string ("scheme://host:port"). Building a proxy is
+	// cheap, but reusing one preserves the underlying http.Transport connection
+	// pool, which matters for throughput.
+	proxyCache sync.Map // map[string]*httputil.ReverseProxy
+
+	// tlsCache holds pre-parsed *tls.Config values keyed by the concatenation
+	// of the PEM certificate and key. Parsing a keypair is relatively expensive
+	// and the same cert is likely reused across many connections.
+	tlsCache sync.Map // map[string]*tls.Config
+}
+
+// ---------------------------------------------------------------------------
+// chanListener – net.Listener backed by a channel
+// ---------------------------------------------------------------------------
+
+// chanListener implements net.Listener by receiving net.Conn values over a
+// buffered channel. This lets the netstack TCP forwarder hand off connections
+// directly to a running http.Server without any real OS socket.
+type chanListener struct {
+	connCh chan net.Conn
+	closed chan struct{}
+	once   sync.Once
+}
+
+func newChanListener() *chanListener {
+	return &chanListener{
+		connCh: make(chan net.Conn, 128),
+		closed: make(chan struct{}),
+	}
+}
+
+// Accept blocks until a connection is available or the listener is closed.
+func (l *chanListener) Accept() (net.Conn, error) {
+	select {
+	case conn, ok := <-l.connCh:
+		if !ok {
+			return nil, net.ErrClosed
+		}
+		return conn, nil
+	case <-l.closed:
+		return nil, net.ErrClosed
+	}
+}
+
+// Close shuts down the listener; subsequent Accept calls return net.ErrClosed.
+func (l *chanListener) Close() error {
+	l.once.Do(func() { close(l.closed) })
+	return nil
+}
+
+// Addr returns a placeholder address (the listener has no real OS socket).
+func (l *chanListener) Addr() net.Addr {
+	return &net.TCPAddr{}
+}
+
+// send delivers conn to the listener. Returns false if the listener is already
+// closed, in which case the caller is responsible for closing conn.
+func (l *chanListener) send(conn net.Conn) bool {
+	select {
+	case l.connCh <- conn:
+		return true
+	case <-l.closed:
+		return false
+	}
+}
+
+// ---------------------------------------------------------------------------
+// httpConnCtx – conn wrapper that carries a SubnetRule through the listener
+// ---------------------------------------------------------------------------
+
+// httpConnCtx wraps a net.Conn so the matching SubnetRule and TLS state can
+// be passed through the chanListener into the http.Server's ConnContext
+// callback, making them available to request handlers via the request context.
+type httpConnCtx struct {
+	net.Conn
+	rule  *SubnetRule
+	isTLS bool // true when the conn was wrapped with tls.Server
+}
+
+// connCtxKey is the unexported context key used to store a *SubnetRule on the
+// per-connection context created by http.Server.ConnContext.
+type connCtxKey struct{}
+
+// connTLSKey is the unexported context key used to store the isTLS flag on
+// the per-connection context created by http.Server.ConnContext.
+type connTLSKey struct{}
+
+// ---------------------------------------------------------------------------
+// Constructor and lifecycle
+// ---------------------------------------------------------------------------
+
+// NewHTTPHandler creates an HTTPHandler attached to the given stack and
+// ProxyHandler. Call Start to begin serving connections.
+func NewHTTPHandler(s *stack.Stack, ph *ProxyHandler) *HTTPHandler {
+	return &HTTPHandler{
+		stack:        s,
+		proxyHandler: ph,
+	}
+}
+
+// SetRequestLogger attaches an HTTPRequestLogger so that every proxied request
+// is recorded and periodically shipped to the server.
+func (h *HTTPHandler) SetRequestLogger(rl *HTTPRequestLogger) {
+	h.requestLogger = rl
+}
+
+// Start launches the internal http.Server that services connections delivered
+// via HandleConn. The server runs for the lifetime of the HTTPHandler; call
+// Close to stop it.
+func (h *HTTPHandler) Start() error {
+	h.listener = newChanListener()
+
+	h.server = &http.Server{
+		Handler: http.HandlerFunc(h.handleRequest),
+		// ConnContext runs once per accepted connection and attaches the
+		// SubnetRule carried by httpConnCtx to the connection's context so
+		// that handleRequest can retrieve it without any global state.
+		ConnContext: func(ctx context.Context, c net.Conn) context.Context {
+			if cc, ok := c.(*httpConnCtx); ok {
+				ctx = context.WithValue(ctx, connCtxKey{}, cc.rule)
+				ctx = context.WithValue(ctx, connTLSKey{}, cc.isTLS)
+			}
+			return ctx
+		},
+	}
+
+	go func() {
+		if err := h.server.Serve(h.listener); err != nil && err != http.ErrServerClosed {
+			logger.Error("HTTP handler: server exited unexpectedly: %v", err)
+		}
+	}()
+
+	logger.Debug("HTTP handler: ready — routing determined per SubnetRule on ports 80/443")
+	return nil
+}
+
+// HandleConn accepts a TCP connection from the netstack forwarder together
+// with the SubnetRule that matched it. The HTTP handler takes full ownership
+// of the connection's lifecycle; the caller must NOT close conn after this call.
+//
+// When rule.Protocol is "https", TLS termination is performed on conn using
+// the certificate and key stored in rule.TLSCert and rule.TLSKey before the
+// connection is passed to the HTTP server. The HTTP server itself is always
+// plain-HTTP; TLS is fully unwrapped at this layer.
+func (h *HTTPHandler) HandleConn(conn net.Conn, rule *SubnetRule) {
+	var effectiveConn net.Conn = conn
+
+	if rule.Protocol == "https" {
+		// Only perform TLS termination for connections arriving on port 443.
+		// Connections on port 80 are passed through as plain HTTP so that
+		// handleRequest can issue the HTTP→HTTPS redirect.
+		doTLS := false
+		if tcpAddr, ok := conn.LocalAddr().(*net.TCPAddr); ok {
+			doTLS = tcpAddr.Port == 443
+		}
+		if doTLS {
+			tlsCfg, err := h.getTLSConfig(rule)
+			if err != nil {
+				logger.Error("HTTP handler: cannot build TLS config for connection from %s: %v",
+					conn.RemoteAddr(), err)
+				conn.Close()
+				return
+			}
+			// tls.Server wraps the raw conn; the TLS handshake is deferred until
+			// the first Read, which the http.Server will trigger naturally.
+			effectiveConn = tls.Server(conn, tlsCfg)
+		}
+	}
+
+	wrapped := &httpConnCtx{Conn: effectiveConn, rule: rule, isTLS: effectiveConn != conn}
+	if !h.listener.send(wrapped) {
+		// Listener is already closed — clean up the orphaned connection.
+		effectiveConn.Close()
+	}
+}
+
+// Close gracefully shuts down the HTTP server and the underlying channel
+// listener, causing the goroutine started in Start to exit.
+func (h *HTTPHandler) Close() error {
+	if h.server != nil {
+		if err := h.server.Close(); err != nil {
+			return err
+		}
+	}
+	if h.listener != nil {
+		h.listener.Close()
+	}
+	return nil
+}
+
+// ---------------------------------------------------------------------------
+// Internal helpers
+// ---------------------------------------------------------------------------
+
+// getTLSConfig returns a *tls.Config for the cert/key pair in rule, using a
+// cache to avoid re-parsing the same keypair on every connection.
+// The cache key is the concatenation of the PEM cert and key strings, so
+// different rules that happen to share the same material hit the same entry.
+func (h *HTTPHandler) getTLSConfig(rule *SubnetRule) (*tls.Config, error) {
+	cacheKey := rule.TLSCert + "|" + rule.TLSKey
+	if v, ok := h.tlsCache.Load(cacheKey); ok {
+		return v.(*tls.Config), nil
+	}
+
+	cert, err := tls.X509KeyPair([]byte(rule.TLSCert), []byte(rule.TLSKey))
+	if err != nil {
+		return nil, fmt.Errorf("failed to parse TLS keypair: %w", err)
+	}
+	cfg := &tls.Config{
+		Certificates: []tls.Certificate{cert},
+	}
+	// LoadOrStore is safe under concurrent calls: if two goroutines race here
+	// both will produce a valid config; the loser's work is discarded.
+	actual, _ := h.tlsCache.LoadOrStore(cacheKey, cfg)
+	return actual.(*tls.Config), nil
+}
+
+// getProxy returns a cached *httputil.ReverseProxy for the given target,
+// creating one on first use. Reusing the proxy preserves its http.Transport
+// connection pool, avoiding repeated TCP/TLS handshakes to the downstream.
+func (h *HTTPHandler) getProxy(target HTTPTarget) *httputil.ReverseProxy {
+	scheme := target.Scheme
+	cacheKey := fmt.Sprintf("%s://%s:%d", scheme, target.DestAddr, target.DestPort)
+
+	if v, ok := h.proxyCache.Load(cacheKey); ok {
+		return v.(*httputil.ReverseProxy)
+	}
+
+	targetURL := &url.URL{
+		Scheme: scheme,
+		Host:   fmt.Sprintf("%s:%d", target.DestAddr, target.DestPort),
+	}
+	var transport http.RoundTripper = http.DefaultTransport
+	if target.Scheme == "https" {
+		// Allow self-signed certificates on downstream HTTPS targets.
+		transport = &http.Transport{
+			TLSClientConfig: &tls.Config{
+				InsecureSkipVerify: true, //nolint:gosec // downstream self-signed certs are a supported configuration
+			},
+		}
+	}
+
+	proxy := &httputil.ReverseProxy{
+		Rewrite: func(pr *httputil.ProxyRequest) {
+			pr.SetURL(targetURL)
+			if host := pr.In.Host; host != "" {
+				pr.Out.Host = host
+			}
+			// SetXForwarded sets X-Forwarded-For from the inbound request's
+			// RemoteAddr (the WireGuard/netstack client address), along with
+			// X-Forwarded-Host and X-Forwarded-Proto. Using Rewrite instead of
+			// Director means the proxy does not append its own automatic
+			// X-Forwarded-For entry, so the header is set exactly once.
+			pr.SetXForwarded()
+
+			// SetXForwarded derives X-Forwarded-Proto from pr.In.TLS,
+			// which is nil because httpConnCtx wraps *tls.Conn behind
+			// net.Conn. Override using the context flag set by ConnContext.
+			if isTLS, _ := pr.In.Context().Value(connTLSKey{}).(bool); isTLS {
+				pr.Out.Header.Set("X-Forwarded-Proto", "https")
+			}
+		},
+		Transport: transport,
+	}
+
+	proxy.ErrorHandler = func(w http.ResponseWriter, r *http.Request, err error) {
+		logger.Error("HTTP handler: upstream error (%s %s -> %s): %v",
+			r.Method, r.URL.RequestURI(), cacheKey, err)
+		http.Error(w, "Bad Gateway", http.StatusBadGateway)
+	}
+
+	actual, _ := h.proxyCache.LoadOrStore(cacheKey, proxy)
+	return actual.(*httputil.ReverseProxy)
+}
+
+// statusCapture wraps an http.ResponseWriter and records the HTTP status code
+// written by the upstream handler. If WriteHeader is never called the status
+// defaults to 200 (http.StatusOK), matching net/http semantics.
+type statusCapture struct {
+	http.ResponseWriter
+	status int
+}
+
+func (sc *statusCapture) WriteHeader(code int) {
+	sc.status = code
+	sc.ResponseWriter.WriteHeader(code)
+}
+
+func (sc *statusCapture) Unwrap() http.ResponseWriter {
+	return sc.ResponseWriter
+}
+
+func (sc *statusCapture) Flush() {
+	if flusher, ok := sc.ResponseWriter.(http.Flusher); ok {
+		flusher.Flush()
+	}
+}
+
+func (sc *statusCapture) Hijack() (net.Conn, *bufio.ReadWriter, error) {
+	hijacker, ok := sc.ResponseWriter.(http.Hijacker)
+	if !ok {
+		return nil, nil, errors.New("underlying response writer does not support hijacking")
+	}
+	return hijacker.Hijack()
+}
+
+// handleRequest is the http.Handler entry point. It retrieves the SubnetRule
+// attached to the connection by ConnContext, selects the first configured
+// downstream target, and forwards the request via the cached ReverseProxy.
+//
+// TODO: add host/path-based routing across multiple HTTPTargets once the
+// configuration model evolves beyond a single target per rule.
+func (h *HTTPHandler) handleRequest(w http.ResponseWriter, r *http.Request) {
+	rule, _ := r.Context().Value(connCtxKey{}).(*SubnetRule)
+	if rule == nil || len(rule.HTTPTargets) == 0 {
+		logger.Error("HTTP handler: no downstream targets for request %s %s", r.Method, r.URL.RequestURI())
+		http.Error(w, "no targets configured", http.StatusBadGateway)
+		return
+	}
+
+	// If the rule is HTTPS but the incoming request arrived over plain HTTP
+	// (port 80), redirect to HTTPS. We use the isTLS flag stored on the
+	// connection context rather than r.TLS, because Go's http.Server calls
+	// ConnectionState() before the TLS handshake completes, so r.TLS.Version
+	// is 0 even for genuine TLS connections at that point.
+	isTLS, _ := r.Context().Value(connTLSKey{}).(bool)
+	if rule.Protocol == "https" && !isTLS {
+		host := r.Host
+		if host == "" {
+			host = r.URL.Host
+		}
+		httpsURL := "https://" + host + r.RequestURI
+		logger.Info("HTTP handler: redirecting %s %s -> %s (TLS cert present)", r.Method, r.URL.RequestURI(), httpsURL)
+		http.Redirect(w, r, httpsURL, http.StatusPermanentRedirect)
+		return
+	}
+
+	target := rule.HTTPTargets[0]
+	scheme := target.Scheme
+	logger.Info("HTTP handler: %s %s -> %s://%s:%d",
+		r.Method, r.URL.RequestURI(), scheme, target.DestAddr, target.DestPort)
+
+	timestamp := time.Now()
+	sc := &statusCapture{ResponseWriter: w, status: http.StatusOK}
+
+	h.getProxy(target).ServeHTTP(sc, r)
+
+	if h.requestLogger != nil && rule.ResourceId != 0 {
+		h.requestLogger.LogRequest(HTTPRequestLog{
+			ResourceID: rule.ResourceId,
+			Timestamp:  timestamp,
+			Method:     r.Method,
+			Scheme:     rule.Protocol,
+			Host:       r.Host,
+			Path:       r.URL.Path,
+			RawQuery:   r.URL.RawQuery,
+			UserAgent:  r.UserAgent(),
+			SourceAddr: r.RemoteAddr,
+			TLS:        rule.Protocol == "https",
+		})
+	}
+}

netstack2/http_handler_test.go

@@ -0,0 +1,97 @@
+package netstack2
+
+import (
+	"context"
+	"net"
+	"net/http"
+	"net/http/httptest"
+	"net/url"
+	"testing"
+
+	"github.com/gorilla/websocket"
+)
+
+func TestHTTPHandlerProxiesWebSocketUpgrade(t *testing.T) {
+	upgrader := websocket.Upgrader{}
+	backend := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		conn, err := upgrader.Upgrade(w, r, nil)
+		if err != nil {
+			t.Errorf("upgrade failed: %v", err)
+			return
+		}
+		defer conn.Close()
+
+		messageType, payload, err := conn.ReadMessage()
+		if err != nil {
+			t.Errorf("read failed: %v", err)
+			return
+		}
+		if err := conn.WriteMessage(messageType, append([]byte("echo:"), payload...)); err != nil {
+			t.Errorf("write failed: %v", err)
+		}
+	}))
+	defer backend.Close()
+
+	backendURL, err := url.Parse(backend.URL)
+	if err != nil {
+		t.Fatalf("parse backend URL: %v", err)
+	}
+	backendHost, backendPort, err := net.SplitHostPort(backendURL.Host)
+	if err != nil {
+		t.Fatalf("split backend host: %v", err)
+	}
+	port, err := net.LookupPort("tcp", backendPort)
+	if err != nil {
+		t.Fatalf("parse backend port: %v", err)
+	}
+
+	handler := NewHTTPHandler(nil, nil)
+	rule := &SubnetRule{
+		Protocol: "http",
+		HTTPTargets: []HTTPTarget{
+			{
+				DestAddr: backendHost,
+				DestPort: uint16(port),
+				Scheme:   backendURL.Scheme,
+			},
+		},
+	}
+
+	frontend := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		ctx := context.WithValue(r.Context(), connCtxKey{}, rule)
+		handler.handleRequest(w, r.WithContext(ctx))
+	}))
+	defer frontend.Close()
+
+	frontendURL, err := url.Parse(frontend.URL)
+	if err != nil {
+		t.Fatalf("parse frontend URL: %v", err)
+	}
+	wsURL := url.URL{
+		Scheme:   "ws",
+		Host:     frontendURL.Host,
+		Path:     "/socket",
+		RawQuery: "token=test",
+	}
+
+	conn, _, err := websocket.DefaultDialer.Dial(wsURL.String(), nil)
+	if err != nil {
+		t.Fatalf("dial websocket through proxy: %v", err)
+	}
+	defer conn.Close()
+
+	if err := conn.WriteMessage(websocket.TextMessage, []byte("hello")); err != nil {
+		t.Fatalf("write websocket message: %v", err)
+	}
+
+	messageType, payload, err := conn.ReadMessage()
+	if err != nil {
+		t.Fatalf("read websocket message: %v", err)
+	}
+	if messageType != websocket.TextMessage {
+		t.Fatalf("message type = %d, want %d", messageType, websocket.TextMessage)
+	}
+	if got, want := string(payload), "echo:hello"; got != want {
+		t.Fatalf("payload = %q, want %q", got, want)
+	}
+}

netstack2/http_request_log.go

@@ -0,0 +1,175 @@
+package netstack2
+
+import (
+	"bytes"
+	"compress/zlib"
+	"encoding/base64"
+	"encoding/json"
+	"sync"
+	"time"
+
+	"github.com/fosrl/newt/logger"
+)
+
+// HTTPRequestLog represents a single HTTP/HTTPS request proxied through the handler.
+type HTTPRequestLog struct {
+	RequestID  string    `json:"requestId"`
+	ResourceID int       `json:"resourceId"`
+	Timestamp  time.Time `json:"timestamp"`
+	Method     string    `json:"method"`
+	Scheme     string    `json:"scheme"`
+	Host       string    `json:"host"`
+	Path       string    `json:"path"`
+	RawQuery   string    `json:"rawQuery,omitempty"`
+	UserAgent  string    `json:"userAgent,omitempty"`
+	SourceAddr string    `json:"sourceAddr"`
+	TLS        bool      `json:"tls"`
+}
+
+// HTTPRequestLogger buffers HTTP request logs and periodically flushes them
+// to the server via a configurable SendFunc.
+type HTTPRequestLogger struct {
+	mu        sync.Mutex
+	pending   []HTTPRequestLog
+	sendFn    SendFunc
+	stopCh    chan struct{}
+	flushDone chan struct{}
+}
+
+// NewHTTPRequestLogger creates a new HTTPRequestLogger and starts its background flush loop.
+func NewHTTPRequestLogger() *HTTPRequestLogger {
+	rl := &HTTPRequestLogger{
+		pending:   make([]HTTPRequestLog, 0),
+		stopCh:    make(chan struct{}),
+		flushDone: make(chan struct{}),
+	}
+	go rl.backgroundLoop()
+	return rl
+}
+
+// SetSendFunc sets the callback used to send compressed HTTP request log batches
+// to the server. This can be called after construction once the websocket
+// client is available.
+func (rl *HTTPRequestLogger) SetSendFunc(fn SendFunc) {
+	rl.mu.Lock()
+	defer rl.mu.Unlock()
+	rl.sendFn = fn
+}
+
+// LogRequest adds an HTTP request log entry to the buffer. If the buffer
+// reaches maxBufferedSessions entries a flush is triggered immediately.
+func (rl *HTTPRequestLogger) LogRequest(log HTTPRequestLog) {
+	if log.RequestID == "" {
+		log.RequestID = generateSessionID()
+	}
+
+	rl.mu.Lock()
+	rl.pending = append(rl.pending, log)
+	shouldFlush := len(rl.pending) >= maxBufferedSessions
+	rl.mu.Unlock()
+
+	if shouldFlush {
+		rl.flush()
+	}
+}
+
+// backgroundLoop handles periodic flushing of buffered request logs.
+func (rl *HTTPRequestLogger) backgroundLoop() {
+	defer close(rl.flushDone)
+
+	ticker := time.NewTicker(flushInterval)
+	defer ticker.Stop()
+
+	for {
+		select {
+		case <-rl.stopCh:
+			return
+		case <-ticker.C:
+			rl.flush()
+		}
+	}
+}
+
+// flush drains the pending buffer, compresses with zlib, and sends via the SendFunc.
+// On send failure the batch is re-queued, capped at maxBufferedSessions*5 entries
+// to prevent unbounded memory growth when the server is unreachable.
+func (rl *HTTPRequestLogger) flush() {
+	rl.mu.Lock()
+	if len(rl.pending) == 0 {
+		rl.mu.Unlock()
+		return
+	}
+	batch := rl.pending
+	rl.pending = make([]HTTPRequestLog, 0)
+	sendFn := rl.sendFn
+	rl.mu.Unlock()
+
+	if sendFn == nil {
+		logger.Debug("HTTP request logger: no send function configured, discarding %d requests", len(batch))
+		return
+	}
+
+	compressed, err := compressRequestLogs(batch)
+	if err != nil {
+		logger.Error("HTTP request logger: failed to compress %d requests: %v", len(batch), err)
+		return
+	}
+
+	if err := sendFn(compressed); err != nil {
+		logger.Error("HTTP request logger: failed to send %d requests: %v", len(batch), err)
+		// Re-queue the batch so we don't lose data
+		rl.mu.Lock()
+		rl.pending = append(batch, rl.pending...)
+		// Cap re-queued data to prevent unbounded growth if server is unreachable
+		if len(rl.pending) > maxBufferedSessions*5 {
+			dropped := len(rl.pending) - maxBufferedSessions*5
+			rl.pending = rl.pending[:maxBufferedSessions*5]
+			logger.Warn("HTTP request logger: buffer overflow, dropped %d oldest requests", dropped)
+		}
+		rl.mu.Unlock()
+		return
+	}
+
+	logger.Info("HTTP request logger: sent %d requests to server", len(batch))
+}
+
+// compressRequestLogs JSON-encodes the request logs, compresses with zlib, and
+// returns a base64-encoded string suitable for embedding in a JSON message.
+func compressRequestLogs(logs []HTTPRequestLog) (string, error) {
+	jsonData, err := json.Marshal(logs)
+	if err != nil {
+		return "", err
+	}
+
+	var buf bytes.Buffer
+	w, err := zlib.NewWriterLevel(&buf, zlib.BestCompression)
+	if err != nil {
+		return "", err
+	}
+	if _, err := w.Write(jsonData); err != nil {
+		w.Close()
+		return "", err
+	}
+	if err := w.Close(); err != nil {
+		return "", err
+	}
+
+	return base64.StdEncoding.EncodeToString(buf.Bytes()), nil
+}
+
+// Close shuts down the background loop and performs one final flush to send
+// any remaining buffered requests to the server.
+func (rl *HTTPRequestLogger) Close() {
+	select {
+	case <-rl.stopCh:
+		// Already closed
+		return
+	default:
+		close(rl.stopCh)
+	}
+
+	// Wait for the background loop to exit so we don't race on flush
+	<-rl.flushDone
+
+	rl.flush()
+}

netstack2/proxy.go

@@ -0,0 +1,899 @@
+package netstack2
+
+import (
+	"context"
+	"fmt"
+	"net"
+	"net/netip"
+	"sync"
+	"sync/atomic"
+	"time"
+
+	"github.com/fosrl/newt/logger"
+	"gvisor.dev/gvisor/pkg/buffer"
+	"gvisor.dev/gvisor/pkg/tcpip"
+	"gvisor.dev/gvisor/pkg/tcpip/checksum"
+	"gvisor.dev/gvisor/pkg/tcpip/header"
+	"gvisor.dev/gvisor/pkg/tcpip/link/channel"
+	"gvisor.dev/gvisor/pkg/tcpip/network/ipv4"
+	"gvisor.dev/gvisor/pkg/tcpip/network/ipv6"
+	"gvisor.dev/gvisor/pkg/tcpip/stack"
+	"gvisor.dev/gvisor/pkg/tcpip/transport/icmp"
+	"gvisor.dev/gvisor/pkg/tcpip/transport/tcp"
+	"gvisor.dev/gvisor/pkg/tcpip/transport/udp"
+)
+
+const (
+	// udpAccessSessionTimeout is how long a UDP access session stays alive without traffic
+	// before being considered ended by the access logger
+	udpAccessSessionTimeout = 120 * time.Second
+)
+
+// PortRange represents an allowed range of ports (inclusive) with optional protocol filtering
+// Protocol can be "tcp", "udp", or "" (empty string means both protocols)
+type PortRange struct {
+	Min      uint16
+	Max      uint16
+	Protocol string // "tcp", "udp", or "" for both
+}
+
+// SubnetRule represents a subnet with optional port restrictions and source address
+// When RewriteTo is set, DNAT (Destination Network Address Translation) is performed:
+//   - Incoming packets: destination IP is rewritten to the resolved RewriteTo address
+//   - Outgoing packets: source IP is rewritten back to the original destination
+//
+// RewriteTo can be either:
+//   - An IP address with CIDR notation (e.g., "192.168.1.1/32")
+//   - A domain name (e.g., "example.com") which will be resolved at request time
+//
+// This allows transparent proxying where traffic appears to come from the rewritten address
+type SubnetRule struct {
+	SourcePrefix netip.Prefix // Source IP prefix (who is sending)
+	DestPrefix   netip.Prefix // Destination IP prefix (where it's going)
+	DisableIcmp  bool         // If true, ICMP traffic is blocked for this subnet
+	RewriteTo    string       // Optional rewrite address for DNAT - can be IP/CIDR or domain name
+	PortRanges   []PortRange  // empty slice means all ports allowed
+	ResourceId   int          // Optional resource ID from the server for access logging
+
+	// HTTP proxy configuration (optional).
+	// When Protocol is non-empty the TCP connection is handled by HTTPHandler
+	// instead of the raw TCP forwarder.
+	Protocol    string       // "", "http", or "https" — controls the incoming (client-facing) protocol
+	HTTPTargets []HTTPTarget // downstream services to proxy requests to
+	TLSCert     string       // PEM-encoded certificate for incoming HTTPS termination
+	TLSKey      string       // PEM-encoded private key for incoming HTTPS termination
+}
+
+// GetAllRules returns a copy of all subnet rules
+func (sl *SubnetLookup) GetAllRules() []SubnetRule {
+	sl.mu.RLock()
+	defer sl.mu.RUnlock()
+
+	var rules []SubnetRule
+	for _, destTriePtr := range sl.sourceTrie.All() {
+		if destTriePtr == nil {
+			continue
+		}
+		for _, rule := range destTriePtr.rules {
+			rules = append(rules, *rule)
+		}
+	}
+	return rules
+}
+
+// connKey uniquely identifies a connection for NAT tracking
+type connKey struct {
+	srcIP   string
+	srcPort uint16
+	dstIP   string
+	dstPort uint16
+	proto   uint8
+}
+
+// reverseConnKey uniquely identifies a connection for reverse NAT lookup (reply direction)
+// Key structure: (rewrittenTo, originalSrcIP, originalSrcPort, originalDstPort, proto)
+// This allows O(1) lookup of NAT entries for reply packets
+type reverseConnKey struct {
+	rewrittenTo     string // The address we rewrote to (becomes src in replies)
+	originalSrcIP   string // Original source IP (becomes dst in replies)
+	originalSrcPort uint16 // Original source port (becomes dst port in replies)
+	originalDstPort uint16 // Original destination port (becomes src port in replies)
+	proto           uint8
+}
+
+// destKey identifies a destination for handler lookups (without source port since it may change)
+type destKey struct {
+	srcIP   string
+	dstIP   string
+	dstPort uint16
+	proto   uint8
+}
+
+// natState tracks NAT translation state for reverse translation
+type natState struct {
+	originalDst netip.Addr // Original destination before DNAT
+	rewrittenTo netip.Addr // The address we rewrote to
+}
+
+// ProxyHandler handles packet injection and extraction for promiscuous mode
+type ProxyHandler struct {
+	proxyStack        *stack.Stack
+	proxyEp           *channel.Endpoint
+	proxyNotifyHandle *channel.NotificationHandle
+	tcpHandler        *TCPHandler
+	udpHandler        *UDPHandler
+	icmpHandler       *ICMPHandler
+	httpHandler       *HTTPHandler
+	subnetLookup      *SubnetLookup
+	natTable          map[connKey]*natState
+	reverseNatTable   map[reverseConnKey]*natState // Reverse lookup map for O(1) reply packet NAT
+	destRewriteTable  map[destKey]netip.Addr       // Maps original dest to rewritten dest for handler lookups
+	resourceTable     map[destKey]int              // Maps connection key to resource ID for access logging
+	natMu             sync.RWMutex
+	enabled           bool
+	icmpReplies       chan []byte          // Channel for ICMP reply packets to be sent back through the tunnel
+	notifiable        channel.Notification // Notification handler for triggering reads
+	accessLogger      *AccessLogger        // Access logger for tracking sessions
+	httpRequestLogger *HTTPRequestLogger   // HTTP request logger for proxied HTTP/HTTPS requests
+	blocked           atomic.Bool          // when true, all new connections are dropped
+}
+
+// ProxyHandlerOptions configures the proxy handler
+type ProxyHandlerOptions struct {
+	EnableTCP  bool
+	EnableUDP  bool
+	EnableICMP bool
+	MTU        int
+}
+
+// NewProxyHandler creates a new proxy handler for promiscuous mode
+func NewProxyHandler(options ProxyHandlerOptions) (*ProxyHandler, error) {
+	if !options.EnableTCP && !options.EnableUDP && !options.EnableICMP {
+		return nil, nil // No proxy needed
+	}
+
+	handler := &ProxyHandler{
+		enabled:          true,
+		subnetLookup:     NewSubnetLookup(),
+		natTable:         make(map[connKey]*natState),
+		reverseNatTable:  make(map[reverseConnKey]*natState),
+		destRewriteTable: make(map[destKey]netip.Addr),
+		resourceTable:    make(map[destKey]int),
+		icmpReplies:      make(chan []byte, 256), // Buffer for ICMP reply packets
+		accessLogger:     NewAccessLogger(udpAccessSessionTimeout),
+		proxyEp:          channel.New(1024, uint32(options.MTU), ""),
+		proxyStack: stack.New(stack.Options{
+			NetworkProtocols: []stack.NetworkProtocolFactory{
+				ipv4.NewProtocol,
+				ipv6.NewProtocol,
+			},
+			TransportProtocols: []stack.TransportProtocolFactory{
+				tcp.NewProtocol,
+				udp.NewProtocol,
+				icmp.NewProtocol4,
+				icmp.NewProtocol6,
+			},
+		}),
+	}
+
+	// Initialize TCP handler if enabled. The HTTP handler piggybacks on the
+	// TCP forwarder — TCPHandler.handleTCPConn checks the subnet rule for
+	// ports 80/443 and routes matching connections to the HTTP handler, so
+	// the HTTP handler is always initialised alongside TCP.
+	if options.EnableTCP {
+		handler.tcpHandler = NewTCPHandler(handler.proxyStack, handler)
+		if err := handler.tcpHandler.InstallTCPHandler(); err != nil {
+			return nil, fmt.Errorf("failed to install TCP handler: %v", err)
+		}
+
+		handler.httpHandler = NewHTTPHandler(handler.proxyStack, handler)
+		if err := handler.httpHandler.Start(); err != nil {
+			return nil, fmt.Errorf("failed to start HTTP handler: %v", err)
+		}
+
+		handler.httpRequestLogger = NewHTTPRequestLogger()
+		handler.httpHandler.SetRequestLogger(handler.httpRequestLogger)
+		logger.Debug("ProxyHandler: HTTP handler enabled")
+	}
+
+	// Initialize UDP handler if enabled
+	if options.EnableUDP {
+		handler.udpHandler = NewUDPHandler(handler.proxyStack, handler)
+		if err := handler.udpHandler.InstallUDPHandler(); err != nil {
+			return nil, fmt.Errorf("failed to install UDP handler: %v", err)
+		}
+	}
+
+	// Initialize ICMP handler if enabled
+	if options.EnableICMP {
+		handler.icmpHandler = NewICMPHandler(handler.proxyStack, handler)
+		if err := handler.icmpHandler.InstallICMPHandler(); err != nil {
+			return nil, fmt.Errorf("failed to install ICMP handler: %v", err)
+		}
+		logger.Debug("ProxyHandler: ICMP handler enabled")
+	}
+
+	// // Example 1: Add a rule with no port restrictions (all ports allowed)
+	// // This accepts all traffic FROM 10.0.0.0/24 TO 10.20.20.0/24
+	// sourceSubnet := netip.MustParsePrefix("10.0.0.0/24")
+	// destSubnet := netip.MustParsePrefix("10.20.20.0/24")
+	// handler.AddSubnetRule(sourceSubnet, destSubnet, nil)
+
+	// // Example 2: Add a rule with specific port ranges
+	// // This accepts traffic FROM 10.0.0.5/32 TO 10.20.21.21/32 only on ports 80, 443, and 8000-9000
+	// sourceIP := netip.MustParsePrefix("10.0.0.5/32")
+	// destIP := netip.MustParsePrefix("10.20.21.21/32")
+	// handler.AddSubnetRule(sourceIP, destIP, []PortRange{
+	// 	{Min: 80, Max: 80},
+	// 	{Min: 443, Max: 443},
+	// 	{Min: 8000, Max: 9000},
+	// })
+
+	return handler, nil
+}
+
+// AddSubnetRule adds a subnet rule to the proxy handler.
+// HTTP proxy behaviour is configured via rule.Protocol, rule.HTTPTargets,
+// rule.TLSCert, and rule.TLSKey; leave Protocol empty for raw TCP/UDP.
+func (p *ProxyHandler) AddSubnetRule(rule SubnetRule) {
+	if p == nil || !p.enabled {
+		return
+	}
+	p.subnetLookup.AddSubnet(rule)
+}
+
+// SetBlocked enables or disables connection blocking on this proxy handler.
+// When enabled, all new TCP/UDP connections from the tunnel are dropped immediately.
+func (p *ProxyHandler) SetBlocked(v bool) {
+	if p == nil {
+		return
+	}
+	p.blocked.Store(v)
+	if v {
+		logger.Debug("ProxyHandler: connection blocking enabled")
+	} else {
+		logger.Debug("ProxyHandler: connection blocking disabled")
+	}
+}
+
+// IsBlocked returns true if connection blocking is currently enabled.
+func (p *ProxyHandler) IsBlocked() bool {
+	if p == nil {
+		return false
+	}
+	return p.blocked.Load()
+}
+
+// RemoveSubnetRule removes a subnet from the proxy handler
+func (p *ProxyHandler) RemoveSubnetRule(sourcePrefix, destPrefix netip.Prefix) {
+	if p == nil || !p.enabled {
+		return
+	}
+	p.subnetLookup.RemoveSubnet(sourcePrefix, destPrefix)
+}
+
+// GetAllRules returns all subnet rules from the proxy handler
+func (p *ProxyHandler) GetAllRules() []SubnetRule {
+	if p == nil || !p.enabled {
+		return nil
+	}
+	return p.subnetLookup.GetAllRules()
+}
+
+// LookupResourceId looks up the resource ID for a connection
+// Returns 0 if no resource ID is associated with this connection
+func (p *ProxyHandler) LookupResourceId(srcIP, dstIP string, dstPort uint16, proto uint8) int {
+	if p == nil || !p.enabled {
+		return 0
+	}
+
+	key := destKey{
+		srcIP:   srcIP,
+		dstIP:   dstIP,
+		dstPort: dstPort,
+		proto:   proto,
+	}
+
+	p.natMu.RLock()
+	defer p.natMu.RUnlock()
+
+	return p.resourceTable[key]
+}
+
+// GetAccessLogger returns the access logger for session tracking
+func (p *ProxyHandler) GetAccessLogger() *AccessLogger {
+	if p == nil {
+		return nil
+	}
+	return p.accessLogger
+}
+
+// SetAccessLogSender configures the function used to send compressed access log
+// batches to the server. This should be called once the websocket client is available.
+func (p *ProxyHandler) SetAccessLogSender(fn SendFunc) {
+	if p == nil || !p.enabled || p.accessLogger == nil {
+		return
+	}
+	p.accessLogger.SetSendFunc(fn)
+}
+
+// GetHTTPRequestLogger returns the HTTP request logger.
+func (p *ProxyHandler) GetHTTPRequestLogger() *HTTPRequestLogger {
+	if p == nil {
+		return nil
+	}
+	return p.httpRequestLogger
+}
+
+// SetHTTPRequestLogSender configures the function used to send compressed HTTP
+// request log batches to the server. This should be called once the websocket
+// client is available.
+func (p *ProxyHandler) SetHTTPRequestLogSender(fn SendFunc) {
+	if p == nil || !p.enabled || p.httpRequestLogger == nil {
+		return
+	}
+	p.httpRequestLogger.SetSendFunc(fn)
+}
+
+// LookupDestinationRewrite looks up the rewritten destination for a connection
+// This is used by TCP/UDP handlers to find the actual target address
+func (p *ProxyHandler) LookupDestinationRewrite(srcIP, dstIP string, dstPort uint16, proto uint8) (netip.Addr, bool) {
+	if p == nil || !p.enabled {
+		return netip.Addr{}, false
+	}
+
+	key := destKey{
+		srcIP:   srcIP,
+		dstIP:   dstIP,
+		dstPort: dstPort,
+		proto:   proto,
+	}
+
+	p.natMu.RLock()
+	defer p.natMu.RUnlock()
+
+	addr, ok := p.destRewriteTable[key]
+	return addr, ok
+}
+
+// resolveRewriteAddress resolves a rewrite address which can be either:
+// - An IP address with CIDR notation (e.g., "192.168.1.1/32") - returns the IP directly
+// - A plain IP address (e.g., "192.168.1.1") - returns the IP directly
+// - A domain name (e.g., "example.com") - performs DNS lookup
+func (p *ProxyHandler) resolveRewriteAddress(rewriteTo string) (netip.Addr, error) {
+	logger.Debug("Resolving rewrite address: %s", rewriteTo)
+
+	// First, try to parse as a CIDR prefix (e.g., "192.168.1.1/32")
+	if prefix, err := netip.ParsePrefix(rewriteTo); err == nil {
+		return prefix.Addr(), nil
+	}
+
+	// Try to parse as a plain IP address (e.g., "192.168.1.1")
+	if addr, err := netip.ParseAddr(rewriteTo); err == nil {
+		return addr, nil
+	}
+
+	// Not an IP address, treat as domain name - perform DNS lookup
+	ctx, cancel := context.WithTimeout(context.Background(), 5*time.Second)
+	defer cancel()
+
+	ips, err := net.DefaultResolver.LookupIP(ctx, "ip4", rewriteTo)
+	if err != nil {
+		return netip.Addr{}, fmt.Errorf("failed to resolve domain %s: %w", rewriteTo, err)
+	}
+
+	if len(ips) == 0 {
+		return netip.Addr{}, fmt.Errorf("no IP addresses found for domain %s", rewriteTo)
+	}
+
+	// Use the first resolved IP address
+	ip := ips[0]
+	if ip4 := ip.To4(); ip4 != nil {
+		addr := netip.AddrFrom4([4]byte{ip4[0], ip4[1], ip4[2], ip4[3]})
+		logger.Debug("Resolved %s to %s", rewriteTo, addr)
+		return addr, nil
+	}
+
+	return netip.Addr{}, fmt.Errorf("no IPv4 address found for domain %s", rewriteTo)
+}
+
+// Initialize sets up the promiscuous NIC with the netTun's notification system
+func (p *ProxyHandler) Initialize(notifiable channel.Notification) error {
+	if p == nil || !p.enabled {
+		return nil
+	}
+
+	// Store notifiable for triggering notifications on ICMP replies
+	p.notifiable = notifiable
+
+	// Add notification handler
+	p.proxyNotifyHandle = p.proxyEp.AddNotify(notifiable)
+
+	// Create NIC with promiscuous mode
+	tcpipErr := p.proxyStack.CreateNICWithOptions(1, p.proxyEp, stack.NICOptions{
+		Disabled: false,
+		QDisc:    nil,
+	})
+	if tcpipErr != nil {
+		return fmt.Errorf("CreateNIC (proxy): %v", tcpipErr)
+	}
+
+	// Enable promiscuous mode - accepts packets for any destination IP
+	if tcpipErr := p.proxyStack.SetPromiscuousMode(1, true); tcpipErr != nil {
+		return fmt.Errorf("SetPromiscuousMode: %v", tcpipErr)
+	}
+
+	// Enable spoofing - allows sending packets from any source IP
+	if tcpipErr := p.proxyStack.SetSpoofing(1, true); tcpipErr != nil {
+		return fmt.Errorf("SetSpoofing: %v", tcpipErr)
+	}
+
+	// Add default route
+	p.proxyStack.AddRoute(tcpip.Route{
+		Destination: header.IPv4EmptySubnet,
+		NIC:         1,
+	})
+
+	return nil
+}
+
+// HandleIncomingPacket processes incoming packets and determines if they should
+// be injected into the proxy stack
+func (p *ProxyHandler) HandleIncomingPacket(packet []byte) bool {
+	if p == nil || !p.enabled {
+		return false
+	}
+
+	// Check minimum packet size
+	if len(packet) < header.IPv4MinimumSize {
+		return false
+	}
+
+	// Only handle IPv4 for now
+	if packet[0]>>4 != 4 {
+		return false
+	}
+
+	// Parse IPv4 header
+	ipv4Header := header.IPv4(packet)
+	srcIP := ipv4Header.SourceAddress()
+	dstIP := ipv4Header.DestinationAddress()
+
+	// Convert gvisor tcpip.Address to netip.Addr
+	srcBytes := srcIP.As4()
+	srcAddr := netip.AddrFrom4(srcBytes)
+	dstBytes := dstIP.As4()
+	dstAddr := netip.AddrFrom4(dstBytes)
+
+	// Parse transport layer to get destination port
+	var dstPort uint16
+	protocol := ipv4Header.TransportProtocol()
+	headerLen := int(ipv4Header.HeaderLength())
+
+	// Extract port based on protocol
+	switch protocol {
+	case header.TCPProtocolNumber:
+		if len(packet) < headerLen+header.TCPMinimumSize {
+			return false
+		}
+		tcpHeader := header.TCP(packet[headerLen:])
+		dstPort = tcpHeader.DestinationPort()
+	case header.UDPProtocolNumber:
+		if len(packet) < headerLen+header.UDPMinimumSize {
+			return false
+		}
+		udpHeader := header.UDP(packet[headerLen:])
+		dstPort = udpHeader.DestinationPort()
+	case header.ICMPv4ProtocolNumber:
+		// ICMP doesn't have ports, use port 0 (must match rules with no port restrictions)
+		dstPort = 0
+		logger.Debug("HandleIncomingPacket: ICMP packet from %s to %s", srcAddr, dstAddr)
+	default:
+		// For other protocols, use port 0 (must match rules with no port restrictions)
+		dstPort = 0
+		logger.Debug("HandleIncomingPacket: Unknown protocol %d from %s to %s", protocol, srcAddr, dstAddr)
+	}
+
+	// Check if the source IP, destination IP, port, and protocol match any subnet rule
+	matchedRule := p.subnetLookup.Match(srcAddr, dstAddr, dstPort, protocol)
+	if matchedRule != nil {
+		logger.Debug("HandleIncomingPacket: Matched rule for %s -> %s (proto=%d, port=%d, resourceId=%d)",
+			srcAddr, dstAddr, protocol, dstPort, matchedRule.ResourceId)
+
+		// Store resource ID for connections without DNAT as well
+		if matchedRule.ResourceId != 0 && matchedRule.RewriteTo == "" {
+			dKey := destKey{
+				srcIP:   srcAddr.String(),
+				dstIP:   dstAddr.String(),
+				dstPort: dstPort,
+				proto:   uint8(protocol),
+			}
+			p.natMu.Lock()
+			p.resourceTable[dKey] = matchedRule.ResourceId
+			p.natMu.Unlock()
+		}
+
+		// Check if we need to perform DNAT
+		if matchedRule.RewriteTo != "" {
+			// Create connection tracking key using original destination
+			// This allows us to check if we've already resolved for this connection
+			var srcPort uint16
+			switch protocol {
+			case header.TCPProtocolNumber:
+				tcpHeader := header.TCP(packet[headerLen:])
+				srcPort = tcpHeader.SourcePort()
+			case header.UDPProtocolNumber:
+				udpHeader := header.UDP(packet[headerLen:])
+				srcPort = udpHeader.SourcePort()
+			}
+
+			// Key using original destination to track the connection
+			key := connKey{
+				srcIP:   srcAddr.String(),
+				srcPort: srcPort,
+				dstIP:   dstAddr.String(),
+				dstPort: dstPort,
+				proto:   uint8(protocol),
+			}
+
+			// Key for handler lookups (doesn't include srcPort for flexibility)
+			dKey := destKey{
+				srcIP:   srcAddr.String(),
+				dstIP:   dstAddr.String(),
+				dstPort: dstPort,
+				proto:   uint8(protocol),
+			}
+
+			// Store resource ID for access logging if present
+			if matchedRule.ResourceId != 0 {
+				p.natMu.Lock()
+				p.resourceTable[dKey] = matchedRule.ResourceId
+				p.natMu.Unlock()
+			}
+
+			// Check if we already have a NAT entry for this connection
+			p.natMu.RLock()
+			existingEntry, exists := p.natTable[key]
+			p.natMu.RUnlock()
+
+			var newDst netip.Addr
+			if exists {
+				// Use the previously resolved address for this connection
+				newDst = existingEntry.rewrittenTo
+				logger.Debug("Using existing NAT entry for connection: %s -> %s", dstAddr, newDst)
+			} else {
+				// New connection - resolve the rewrite address
+				var err error
+				newDst, err = p.resolveRewriteAddress(matchedRule.RewriteTo)
+				if err != nil {
+					// Failed to resolve, skip DNAT but still proxy the packet
+					logger.Debug("Failed to resolve rewrite address: %v", err)
+					pkb := stack.NewPacketBuffer(stack.PacketBufferOptions{
+						Payload: buffer.MakeWithData(packet),
+					})
+					p.proxyEp.InjectInbound(header.IPv4ProtocolNumber, pkb)
+					return true
+				}
+
+				// Store NAT state for this connection
+				p.natMu.Lock()
+				natEntry := &natState{
+					originalDst: dstAddr,
+					rewrittenTo: newDst,
+				}
+				p.natTable[key] = natEntry
+
+				// Create reverse lookup key for O(1) reply packet lookups
+				// Key: (rewrittenTo, originalSrcIP, originalSrcPort, originalDstPort, proto)
+				reverseKey := reverseConnKey{
+					rewrittenTo:     newDst.String(),
+					originalSrcIP:   srcAddr.String(),
+					originalSrcPort: srcPort,
+					originalDstPort: dstPort,
+					proto:           uint8(protocol),
+				}
+				p.reverseNatTable[reverseKey] = natEntry
+
+				// Store destination rewrite for handler lookups
+				p.destRewriteTable[dKey] = newDst
+
+				// Also store the resource ID under the rewritten destination key so that
+				// TCP/UDP handlers can find it after DNAT (they see the post-NAT dst IP).
+				if matchedRule.ResourceId != 0 {
+					rewrittenKey := destKey{
+						srcIP:   srcAddr.String(),
+						dstIP:   newDst.String(),
+						dstPort: dstPort,
+						proto:   uint8(protocol),
+					}
+					p.resourceTable[rewrittenKey] = matchedRule.ResourceId
+				}
+				p.natMu.Unlock()
+				logger.Debug("New NAT entry for connection: %s -> %s", dstAddr, newDst)
+			}
+
+			// Check if target is loopback - if so, don't rewrite packet destination
+			// as gVisor will drop martian packets. Instead, the handlers will use
+			// destRewriteTable to find the actual target address.
+			if !newDst.IsLoopback() {
+				// Rewrite the packet only for non-loopback destinations
+				packet = p.rewritePacketDestination(packet, newDst)
+				if packet == nil {
+					return false
+				}
+			} else {
+				logger.Debug("Target is loopback, not rewriting packet - handlers will use rewrite table")
+			}
+		}
+
+		// Inject into proxy stack
+		pkb := stack.NewPacketBuffer(stack.PacketBufferOptions{
+			Payload: buffer.MakeWithData(packet),
+		})
+		p.proxyEp.InjectInbound(header.IPv4ProtocolNumber, pkb)
+		logger.Debug("HandleIncomingPacket: Injected packet into proxy stack (proto=%d)", protocol)
+		return true
+	}
+
+	// logger.Debug("HandleIncomingPacket: No matching rule for %s -> %s (proto=%d, port=%d)",
+	// srcAddr, dstAddr, protocol, dstPort)
+	return false
+}
+
+// rewritePacketDestination rewrites the destination IP in a packet and recalculates checksums
+func (p *ProxyHandler) rewritePacketDestination(packet []byte, newDst netip.Addr) []byte {
+	if len(packet) < header.IPv4MinimumSize {
+		return nil
+	}
+
+	// Make a copy to avoid modifying the original
+	pkt := make([]byte, len(packet))
+	copy(pkt, packet)
+
+	ipv4Header := header.IPv4(pkt)
+	headerLen := int(ipv4Header.HeaderLength())
+
+	// Rewrite destination IP
+	newDstBytes := newDst.As4()
+	newDstAddr := tcpip.AddrFrom4(newDstBytes)
+	ipv4Header.SetDestinationAddress(newDstAddr)
+
+	// Recalculate IP checksum
+	ipv4Header.SetChecksum(0)
+	ipv4Header.SetChecksum(^ipv4Header.CalculateChecksum())
+
+	// Update transport layer checksum if needed
+	protocol := ipv4Header.TransportProtocol()
+	switch protocol {
+	case header.TCPProtocolNumber:
+		if len(pkt) >= headerLen+header.TCPMinimumSize {
+			tcpHeader := header.TCP(pkt[headerLen:])
+			tcpHeader.SetChecksum(0)
+			xsum := header.PseudoHeaderChecksum(
+				header.TCPProtocolNumber,
+				ipv4Header.SourceAddress(),
+				ipv4Header.DestinationAddress(),
+				uint16(len(pkt)-headerLen),
+			)
+			xsum = checksum.Checksum(pkt[headerLen:], xsum)
+			tcpHeader.SetChecksum(^xsum)
+		}
+	case header.UDPProtocolNumber:
+		if len(pkt) >= headerLen+header.UDPMinimumSize {
+			udpHeader := header.UDP(pkt[headerLen:])
+			udpHeader.SetChecksum(0)
+			xsum := header.PseudoHeaderChecksum(
+				header.UDPProtocolNumber,
+				ipv4Header.SourceAddress(),
+				ipv4Header.DestinationAddress(),
+				uint16(len(pkt)-headerLen),
+			)
+			xsum = checksum.Checksum(pkt[headerLen:], xsum)
+			udpHeader.SetChecksum(^xsum)
+		}
+	}
+
+	return pkt
+}
+
+// rewritePacketSource rewrites the source IP in a packet and recalculates checksums (for reverse NAT)
+func (p *ProxyHandler) rewritePacketSource(packet []byte, newSrc netip.Addr) []byte {
+	if len(packet) < header.IPv4MinimumSize {
+		return nil
+	}
+
+	// Make a copy to avoid modifying the original
+	pkt := make([]byte, len(packet))
+	copy(pkt, packet)
+
+	ipv4Header := header.IPv4(pkt)
+	headerLen := int(ipv4Header.HeaderLength())
+
+	// Rewrite source IP
+	newSrcBytes := newSrc.As4()
+	newSrcAddr := tcpip.AddrFrom4(newSrcBytes)
+	ipv4Header.SetSourceAddress(newSrcAddr)
+
+	// Recalculate IP checksum
+	ipv4Header.SetChecksum(0)
+	ipv4Header.SetChecksum(^ipv4Header.CalculateChecksum())
+
+	// Update transport layer checksum if needed
+	protocol := ipv4Header.TransportProtocol()
+	switch protocol {
+	case header.TCPProtocolNumber:
+		if len(pkt) >= headerLen+header.TCPMinimumSize {
+			tcpHeader := header.TCP(pkt[headerLen:])
+			tcpHeader.SetChecksum(0)
+			xsum := header.PseudoHeaderChecksum(
+				header.TCPProtocolNumber,
+				ipv4Header.SourceAddress(),
+				ipv4Header.DestinationAddress(),
+				uint16(len(pkt)-headerLen),
+			)
+			xsum = checksum.Checksum(pkt[headerLen:], xsum)
+			tcpHeader.SetChecksum(^xsum)
+		}
+	case header.UDPProtocolNumber:
+		if len(pkt) >= headerLen+header.UDPMinimumSize {
+			udpHeader := header.UDP(pkt[headerLen:])
+			udpHeader.SetChecksum(0)
+			xsum := header.PseudoHeaderChecksum(
+				header.UDPProtocolNumber,
+				ipv4Header.SourceAddress(),
+				ipv4Header.DestinationAddress(),
+				uint16(len(pkt)-headerLen),
+			)
+			xsum = checksum.Checksum(pkt[headerLen:], xsum)
+			udpHeader.SetChecksum(^xsum)
+		}
+	}
+
+	return pkt
+}
+
+// ReadOutgoingPacket reads packets from the proxy stack that need to be
+// sent back through the tunnel
+func (p *ProxyHandler) ReadOutgoingPacket() *buffer.View {
+	if p == nil || !p.enabled {
+		return nil
+	}
+
+	// First check for ICMP reply packets (non-blocking)
+	select {
+	case icmpReply := <-p.icmpReplies:
+		logger.Debug("ReadOutgoingPacket: Returning ICMP reply packet (%d bytes)", len(icmpReply))
+		return buffer.NewViewWithData(icmpReply)
+	default:
+		// No ICMP reply available, continue to check proxy endpoint
+	}
+
+	pkt := p.proxyEp.Read()
+	if pkt != nil {
+		view := pkt.ToView()
+		pkt.DecRef()
+
+		// Check if we need to perform reverse NAT
+		packet := view.AsSlice()
+		if len(packet) >= header.IPv4MinimumSize && packet[0]>>4 == 4 {
+			ipv4Header := header.IPv4(packet)
+			srcIP := ipv4Header.SourceAddress()
+			dstIP := ipv4Header.DestinationAddress()
+			protocol := ipv4Header.TransportProtocol()
+			headerLen := int(ipv4Header.HeaderLength())
+
+			// Extract ports
+			var srcPort, dstPort uint16
+			switch protocol {
+			case header.TCPProtocolNumber:
+				if len(packet) >= headerLen+header.TCPMinimumSize {
+					tcpHeader := header.TCP(packet[headerLen:])
+					srcPort = tcpHeader.SourcePort()
+					dstPort = tcpHeader.DestinationPort()
+				}
+			case header.UDPProtocolNumber:
+				if len(packet) >= headerLen+header.UDPMinimumSize {
+					udpHeader := header.UDP(packet[headerLen:])
+					srcPort = udpHeader.SourcePort()
+					dstPort = udpHeader.DestinationPort()
+				}
+			case header.ICMPv4ProtocolNumber:
+				// ICMP packets don't need NAT translation in our implementation
+				// since we construct reply packets with the correct addresses
+				logger.Debug("ReadOutgoingPacket: ICMP packet from %s to %s", srcIP, dstIP)
+				return view
+			}
+
+			// Look up NAT state for reverse translation using O(1) reverse lookup map
+			// Key: (rewrittenTo, originalSrcIP, originalSrcPort, originalDstPort, proto)
+			// For reply packets:
+			// - reply's srcIP = rewrittenTo (the address we rewrote to)
+			// - reply's dstIP = originalSrcIP (original source IP)
+			// - reply's srcPort = originalDstPort (original destination port)
+			// - reply's dstPort = originalSrcPort (original source port)
+			p.natMu.RLock()
+			reverseKey := reverseConnKey{
+				rewrittenTo:     srcIP.String(), // Reply's source is the rewritten address
+				originalSrcIP:   dstIP.String(), // Reply's destination is the original source
+				originalSrcPort: dstPort,        // Reply's destination port is the original source port
+				originalDstPort: srcPort,        // Reply's source port is the original destination port
+				proto:           uint8(protocol),
+			}
+			natEntry := p.reverseNatTable[reverseKey]
+			p.natMu.RUnlock()
+
+			if natEntry != nil {
+				// Perform reverse NAT - rewrite source to original destination
+				packet = p.rewritePacketSource(packet, natEntry.originalDst)
+				if packet != nil {
+					return buffer.NewViewWithData(packet)
+				}
+			}
+		}
+
+		return view
+	}
+
+	return nil
+}
+
+// QueueICMPReply queues an ICMP reply packet to be sent back through the tunnel
+func (p *ProxyHandler) QueueICMPReply(packet []byte) bool {
+	if p == nil || !p.enabled {
+		return false
+	}
+
+	select {
+	case p.icmpReplies <- packet:
+		logger.Debug("QueueICMPReply: Queued ICMP reply packet (%d bytes)", len(packet))
+		// Trigger notification so WriteNotify picks up the packet
+		if p.notifiable != nil {
+			p.notifiable.WriteNotify()
+		}
+		return true
+	default:
+		logger.Info("QueueICMPReply: ICMP reply channel full, dropping packet")
+		return false
+	}
+}
+
+// Close cleans up the proxy handler resources
+func (p *ProxyHandler) Close() error {
+	if p == nil || !p.enabled {
+		return nil
+	}
+
+	// Shut down access logger
+	if p.accessLogger != nil {
+		p.accessLogger.Close()
+	}
+
+	// Shut down HTTP request logger
+	if p.httpRequestLogger != nil {
+		p.httpRequestLogger.Close()
+	}
+
+	// Shut down HTTP handler
+	if p.httpHandler != nil {
+		p.httpHandler.Close()
+	}
+
+	// Close ICMP replies channel
+	if p.icmpReplies != nil {
+		close(p.icmpReplies)
+	}
+
+	if p.proxyStack != nil {
+		p.proxyStack.RemoveNIC(1)
+		p.proxyStack.Close()
+	}
+
+	if p.proxyEp != nil {
+		if p.proxyNotifyHandle != nil {
+			p.proxyEp.RemoveNotify(p.proxyNotifyHandle)
+		}
+		p.proxyEp.Close()
+	}
+
+	return nil
+}

netstack2/subnet_lookup.go

@@ -0,0 +1,201 @@
+package netstack2
+
+import (
+	"net/netip"
+	"sync"
+
+	"github.com/gaissmai/bart"
+	"gvisor.dev/gvisor/pkg/tcpip"
+	"gvisor.dev/gvisor/pkg/tcpip/header"
+)
+
+// SubnetLookup provides fast IP subnet and port matching using BART (Binary Aggregated Range Tree)
+// This uses BART Table for O(log n) prefix matching with Supernets() for efficient lookups
+//
+// Architecture:
+// - Two-level BART structure for matching both source AND destination prefixes
+// - Level 1: Source prefix -> Level 2 (destination prefix -> rules)
+// - This reduces search space: only check destination prefixes for matching source prefixes
+type SubnetLookup struct {
+	mu sync.RWMutex
+	// Two-level BART structure:
+	// Level 1: Source prefix -> Level 2 (destination prefix -> rules)
+	// This allows us to first match source prefix, then only check destination prefixes
+	// for matching source prefixes, reducing the search space significantly
+	sourceTrie *bart.Table[*destTrie]
+}
+
+// destTrie is a BART for destination prefixes, containing the actual rules
+type destTrie struct {
+	trie  *bart.Table[[]*SubnetRule]
+	rules []*SubnetRule // All rules for this source prefix (for iteration if needed)
+}
+
+// NewSubnetLookup creates a new subnet lookup table using BART
+func NewSubnetLookup() *SubnetLookup {
+	return &SubnetLookup{
+		sourceTrie: &bart.Table[*destTrie]{},
+	}
+}
+
+// prefixEqual compares two prefixes after masking to handle host bits correctly.
+// For example, 10.0.0.5/24 and 10.0.0.0/24 are treated as equal.
+func prefixEqual(a, b netip.Prefix) bool {
+	return a.Masked() == b.Masked()
+}
+
+// AddSubnet adds a subnet rule to the lookup table.
+// If rule.PortRanges is nil or empty, all ports are allowed.
+// rule.RewriteTo can be either an IP/CIDR (e.g., "192.168.1.1/32") or a domain name (e.g., "example.com").
+// HTTP proxy behaviour is driven by rule.Protocol, rule.HTTPTargets, rule.TLSCert, and rule.TLSKey.
+func (sl *SubnetLookup) AddSubnet(rule SubnetRule) {
+	sl.mu.Lock()
+	defer sl.mu.Unlock()
+
+	rulePtr := &rule
+
+	// Canonicalize source prefix to handle host bits correctly
+	canonicalSourcePrefix := rule.SourcePrefix.Masked()
+
+	// Get or create destination trie for this source prefix
+	destTriePtr, exists := sl.sourceTrie.Get(canonicalSourcePrefix)
+	if !exists {
+		// Create new destination trie for this source prefix
+		destTriePtr = &destTrie{
+			trie:  &bart.Table[[]*SubnetRule]{},
+			rules: make([]*SubnetRule, 0),
+		}
+		sl.sourceTrie.Insert(canonicalSourcePrefix, destTriePtr)
+	}
+
+	// Canonicalize destination prefix to handle host bits correctly
+	// BART masks prefixes internally, so we need to match that behavior in our bookkeeping
+	canonicalDestPrefix := rule.DestPrefix.Masked()
+
+	// Add rule to destination trie
+	// Original behavior: overwrite if same (sourcePrefix, destPrefix) exists
+	// Store as single-element slice to match original overwrite behavior
+	destTriePtr.trie.Insert(canonicalDestPrefix, []*SubnetRule{rulePtr})
+
+	// Update destTriePtr.rules - remove old rule with same canonical prefix if exists, then add new one
+	// Use canonical comparison to handle cases like 10.0.0.5/24 vs 10.0.0.0/24
+	newRules := make([]*SubnetRule, 0, len(destTriePtr.rules)+1)
+	for _, r := range destTriePtr.rules {
+		if !prefixEqual(r.DestPrefix, canonicalDestPrefix) || !prefixEqual(r.SourcePrefix, canonicalSourcePrefix) {
+			newRules = append(newRules, r)
+		}
+	}
+	newRules = append(newRules, rulePtr)
+	destTriePtr.rules = newRules
+}
+
+// RemoveSubnet removes a subnet rule from the lookup table
+func (sl *SubnetLookup) RemoveSubnet(sourcePrefix, destPrefix netip.Prefix) {
+	sl.mu.Lock()
+	defer sl.mu.Unlock()
+
+	// Canonicalize prefixes to handle host bits correctly
+	canonicalSourcePrefix := sourcePrefix.Masked()
+	canonicalDestPrefix := destPrefix.Masked()
+
+	destTriePtr, exists := sl.sourceTrie.Get(canonicalSourcePrefix)
+	if !exists {
+		return
+	}
+
+	// Remove the rule - original behavior: delete exact (sourcePrefix, destPrefix) combination
+	// BART masks prefixes internally, so Delete works with canonical form
+	destTriePtr.trie.Delete(canonicalDestPrefix)
+
+	// Also remove from destTriePtr.rules using canonical comparison
+	// This ensures we remove rules even if they were added with host bits set
+	newDestRules := make([]*SubnetRule, 0, len(destTriePtr.rules))
+	for _, r := range destTriePtr.rules {
+		if !prefixEqual(r.DestPrefix, canonicalDestPrefix) || !prefixEqual(r.SourcePrefix, canonicalSourcePrefix) {
+			newDestRules = append(newDestRules, r)
+		}
+	}
+	destTriePtr.rules = newDestRules
+
+	// Check if the trie is actually empty using BART's Size() method
+	// This is more efficient than iterating and ensures we clean up empty tries
+	// even if there were stale entries in the rules slice (which shouldn't happen
+	// with proper canonicalization, but this provides a definitive check)
+	if destTriePtr.trie.Size() == 0 {
+		sl.sourceTrie.Delete(canonicalSourcePrefix)
+	}
+}
+
+// Match checks if a source IP, destination IP, port, and protocol match any subnet rule
+// Returns the matched rule if ALL of these conditions are met:
+//   - The source IP is in the rule's source prefix
+//   - The destination IP is in the rule's destination prefix
+//   - The port is in an allowed range (or no port restrictions exist)
+//   - The protocol matches (or the port range allows both protocols)
+//
+// proto should be header.TCPProtocolNumber, header.UDPProtocolNumber, or header.ICMPv4ProtocolNumber
+// Returns nil if no rule matches
+// This uses BART's Supernets() for O(log n) prefix matching instead of O(n) iteration
+func (sl *SubnetLookup) Match(srcIP, dstIP netip.Addr, port uint16, proto tcpip.TransportProtocolNumber) *SubnetRule {
+	sl.mu.RLock()
+	defer sl.mu.RUnlock()
+
+	// Convert IP addresses to /32 (IPv4) or /128 (IPv6) prefixes
+	// Supernets() finds all prefixes that contain this IP (i.e., are supernets of /32 or /128)
+	srcPrefix := netip.PrefixFrom(srcIP, srcIP.BitLen())
+	dstPrefix := netip.PrefixFrom(dstIP, dstIP.BitLen())
+
+	// Step 1: Find all source prefixes that contain srcIP using BART's Supernets
+	// This is O(log n) instead of O(n) iteration
+	// Supernets returns all prefixes that are supernets (contain) the given prefix
+	for _, destTriePtr := range sl.sourceTrie.Supernets(srcPrefix) {
+		if destTriePtr == nil {
+			continue
+		}
+
+		// Step 2: Find all destination prefixes that contain dstIP
+		// This is also O(log n) for each matching source prefix
+		for _, rules := range destTriePtr.trie.Supernets(dstPrefix) {
+			if rules == nil {
+				continue
+			}
+
+			// Step 3: Check each rule for ICMP and port restrictions
+			for _, rule := range rules {
+				// Handle ICMP before port range check — ICMP has no ports
+				if proto == header.ICMPv4ProtocolNumber || proto == header.ICMPv6ProtocolNumber {
+					if rule.DisableIcmp {
+						return nil
+					}
+					// ICMP is allowed; port ranges don't apply to ICMP
+					return rule
+				}
+
+				// Check port restrictions
+				if len(rule.PortRanges) == 0 {
+					// No port restrictions, match!
+					return rule
+				}
+
+				// Check if port and protocol are in any of the allowed ranges
+				for _, pr := range rule.PortRanges {
+					if port >= pr.Min && port <= pr.Max {
+						// Check protocol compatibility
+						if pr.Protocol == "" {
+							// Empty protocol means allow both TCP and UDP
+							return rule
+						}
+						// Check if the packet protocol matches the port range protocol
+						if (pr.Protocol == "tcp" && proto == header.TCPProtocolNumber) ||
+							(pr.Protocol == "udp" && proto == header.UDPProtocolNumber) {
+							return rule
+						}
+						// Port matches but protocol doesn't - continue checking other ranges
+					}
+				}
+			}
+		}
+	}
+
+	return nil
+}

network/interface.go

@@ -0,0 +1,169 @@
+package network
+
+import (
+	"fmt"
+	"net"
+	"os/exec"
+	"regexp"
+	"runtime"
+	"strconv"
+	"time"
+
+	"github.com/fosrl/newt/logger"
+	"github.com/vishvananda/netlink"
+)
+
+// ConfigureInterface configures a network interface with an IP address and brings it up
+func ConfigureInterface(interfaceName string, tunnelIp string, mtu int) error {
+	logger.Info("The tunnel IP is: %s", tunnelIp)
+
+	// Parse the IP address and network
+	ip, ipNet, err := net.ParseCIDR(tunnelIp)
+	if err != nil {
+		return fmt.Errorf("invalid IP address: %v", err)
+	}
+
+	// Convert CIDR mask to dotted decimal format (e.g., 255.255.255.0)
+	mask := net.IP(ipNet.Mask).String()
+	destinationAddress := ip.String()
+
+	logger.Debug("The destination address is: %s", destinationAddress)
+
+	// network.SetTunnelRemoteAddress() // what does this do?
+	SetIPv4Settings([]string{destinationAddress}, []string{mask})
+	SetMTU(mtu)
+
+	if interfaceName == "" {
+		return nil
+	}
+
+	switch runtime.GOOS {
+	case "linux":
+		return configureLinux(interfaceName, ip, ipNet)
+	case "darwin":
+		return configureDarwin(interfaceName, ip, ipNet)
+	case "windows":
+		return configureWindows(interfaceName, ip, ipNet)
+	case "android":
+		return nil
+	case "ios":
+		return nil
+	}
+
+	return nil
+}
+
+// waitForInterfaceUp polls the network interface until it's up or times out
+func waitForInterfaceUp(interfaceName string, expectedIP net.IP, timeout time.Duration) error {
+	logger.Info("Waiting for interface %s to be up with IP %s", interfaceName, expectedIP)
+	deadline := time.Now().Add(timeout)
+	pollInterval := 500 * time.Millisecond
+
+	for time.Now().Before(deadline) {
+		// Check if interface exists and is up
+		iface, err := net.InterfaceByName(interfaceName)
+		if err == nil {
+			// Check if interface is up
+			if iface.Flags&net.FlagUp != 0 {
+				// Check if it has the expected IP
+				addrs, err := iface.Addrs()
+				if err == nil {
+					for _, addr := range addrs {
+						ipNet, ok := addr.(*net.IPNet)
+						if ok && ipNet.IP.Equal(expectedIP) {
+							logger.Info("Interface %s is up with correct IP", interfaceName)
+							return nil // Interface is up with correct IP
+						}
+					}
+					logger.Info("Interface %s is up but doesn't have expected IP yet", interfaceName)
+				}
+			} else {
+				logger.Info("Interface %s exists but is not up yet", interfaceName)
+			}
+		} else {
+			logger.Info("Interface %s not found yet: %v", interfaceName, err)
+		}
+
+		// Wait before next check
+		time.Sleep(pollInterval)
+	}
+
+	return fmt.Errorf("timed out waiting for interface %s to be up with IP %s", interfaceName, expectedIP)
+}
+
+func FindUnusedUTUN() (string, error) {
+	ifaces, err := net.Interfaces()
+	if err != nil {
+		return "", fmt.Errorf("failed to list interfaces: %v", err)
+	}
+	used := make(map[int]bool)
+	re := regexp.MustCompile(`^utun(\d+)$`)
+	for _, iface := range ifaces {
+		if matches := re.FindStringSubmatch(iface.Name); len(matches) == 2 {
+			if num, err := strconv.Atoi(matches[1]); err == nil {
+				used[num] = true
+			}
+		}
+	}
+	// Try utun0 up to utun255.
+	for i := 0; i < 256; i++ {
+		if !used[i] {
+			return fmt.Sprintf("utun%d", i), nil
+		}
+	}
+	return "", fmt.Errorf("no unused utun interface found")
+}
+
+func configureDarwin(interfaceName string, ip net.IP, ipNet *net.IPNet) error {
+	logger.Info("Configuring darwin interface: %s", interfaceName)
+
+	prefix, _ := ipNet.Mask.Size()
+	ipStr := fmt.Sprintf("%s/%d", ip.String(), prefix)
+
+	cmd := exec.Command("/sbin/ifconfig", interfaceName, "inet", ipStr, ip.String(), "alias")
+	logger.Info("Running command: %v", cmd)
+
+	out, err := cmd.CombinedOutput()
+	if err != nil {
+		return fmt.Errorf("ifconfig command failed: %v, output: %s", err, out)
+	}
+
+	// Bring up the interface
+	cmd = exec.Command("/sbin/ifconfig", interfaceName, "up")
+	logger.Info("Running command: %v", cmd)
+
+	out, err = cmd.CombinedOutput()
+	if err != nil {
+		return fmt.Errorf("ifconfig up command failed: %v, output: %s", err, out)
+	}
+
+	return nil
+}
+
+func configureLinux(interfaceName string, ip net.IP, ipNet *net.IPNet) error {
+	// Get the interface
+	link, err := netlink.LinkByName(interfaceName)
+	if err != nil {
+		return fmt.Errorf("failed to get interface %s: %v", interfaceName, err)
+	}
+
+	// Create the IP address attributes
+	addr := &netlink.Addr{
+		IPNet: &net.IPNet{
+			IP:   ip,
+			Mask: ipNet.Mask,
+		},
+	}
+
+	// Add the IP address to the interface
+	if err := netlink.AddrAdd(link, addr); err != nil {
+		return fmt.Errorf("failed to add IP address: %v", err)
+	}
+
+	// Bring up the interface
+	if err := netlink.LinkSetUp(link); err != nil {
+		return fmt.Errorf("failed to bring up interface: %v", err)
+	}
+
+	return nil
+}

network/interface_notwindows.go

@@ -0,0 +1,12 @@
+//go:build !windows
+
+package network
+
+import (
+	"fmt"
+	"net"
+)
+
+func configureWindows(interfaceName string, ip net.IP, ipNet *net.IPNet) error {
+	return fmt.Errorf("configureWindows called on non-Windows platform")
+}