github-starred/pangolin
2
0
Fork 0
mirror of https://github.com/fosrl/pangolin.git synced 2026-05-24 03:18:37 -05:00
Code Issues 959 Packages Projects Releases 74 Wiki Activity

[GH-ISSUE #2913] newt holepunch does not re-resolve endpoint hostname after connection loss (stale IP on reconnect) #13072

New Issue
Open
opened 2026-05-13 18:37:11 -05:00 by GiteaMirror · 0 comments
GiteaMirror commented 2026-05-13 18:37:11 -05:00
Owner
Copy Link

Originally created by @timootten on GitHub (Apr 28, 2026).
Original GitHub issue: https://github.com/fosrl/pangolin/issues/2913

Originally assigned to: @oschwartz10612 on GitHub.

Describe the Bug

newt/holepunch does not re-resolve endpoint hostname after connection loss (stale IP on reconnect)

Summary

When the IP address behind the configured gerbil/exit-node hostname changes (e.g. due to a failover, migration, or DNS record update), newt does not perform a fresh DNS lookup on reconnect. Instead, it continues using the IP it resolved at startup, leaving the tunnel in a permanently broken state until newt is manually restarted.

This affects any dynamic IP environment — including setups using tools like external-dns that automatically update DNS records at the provider level during cluster events.

Actual Behavior

Holepunch keeps sending to the originally resolved IP for the entire runtime. No re-resolution occurs even well past the DNS TTL. The only recovery path is manually restarting newt.

Evidence / Logs

Starting hole punch to <exit-node-hostname>
Resolved exit node: <Address B>
Sent UDP hole punch to <Address B>
Sent UDP hole punch to <Address B>
... (repeated indefinitely, no re-resolution to Address A)

No subsequent resolution attempt is visible in the logs after the DNS record changes.

Environment

  • Tool: newt + holepunch
  • DNS managed externally (e.g. Cloudflare via external-dns)
  • Deployment: Kubernetes

Suggested Fix

On each reconnect attempt (or periodically, respecting TTL), re-resolve the exit-node hostname from DNS rather than caching the IP from the initial startup resolution.

Environment

  • OS Type & Version: Debian 13
  • Pangolin Version: 1.17.1
  • Gerbil Version: 1.3.1
  • Traefik Version: 3.6
  • Newt Version: 11.0
  • Olm Version: (if applicable)

To Reproduce

  1. Point the exit-node hostname DNS record to Address A
  2. Start newt — confirm holepunch targets Address A in the logs
  3. Update the DNS record to point to Address B (simulating a failover)
  4. Restart newt — confirm holepunch correctly targets Address B
  5. Update the DNS record back to Address A
  6. Wait longer than the DNS TTL (e.g. 8+ minutes with a TTL of ~60s)
  7. Observe holepunch logs — it continues targeting Address B

Expected Behavior

After the DNS TTL expires, newt/holepunch should re-resolve the exit-node hostname and reconnect using the updated IP address — without requiring a manual restart.

Originally created by @timootten on GitHub (Apr 28, 2026). Original GitHub issue: https://github.com/fosrl/pangolin/issues/2913 Originally assigned to: @oschwartz10612 on GitHub. ### Describe the Bug # newt/holepunch does not re-resolve endpoint hostname after connection loss (stale IP on reconnect) ## Summary When the IP address behind the configured gerbil/exit-node hostname changes (e.g. due to a failover, migration, or DNS record update), `newt` does not perform a fresh DNS lookup on reconnect. Instead, it continues using the IP it resolved at startup, leaving the tunnel in a permanently broken state until `newt` is manually restarted. This affects any dynamic IP environment — including setups using tools like [external-dns](https://github.com/kubernetes-sigs/external-dns) that automatically update DNS records at the provider level during cluster events. ## Actual Behavior Holepunch keeps sending to the originally resolved IP for the entire runtime. No re-resolution occurs even well past the DNS TTL. The only recovery path is manually restarting `newt`. ## Evidence / Logs ``` Starting hole punch to <exit-node-hostname> Resolved exit node: <Address B> Sent UDP hole punch to <Address B> Sent UDP hole punch to <Address B> ... (repeated indefinitely, no re-resolution to Address A) ``` No subsequent resolution attempt is visible in the logs after the DNS record changes. ## Environment - Tool: `newt` + holepunch - DNS managed externally (e.g. Cloudflare via external-dns) - Deployment: Kubernetes ## Suggested Fix On each reconnect attempt (or periodically, respecting TTL), re-resolve the exit-node hostname from DNS rather than caching the IP from the initial startup resolution. ### Environment - OS Type & Version: Debian 13 - Pangolin Version: 1.17.1 - Gerbil Version: 1.3.1 - Traefik Version: 3.6 - Newt Version: 11.0 - Olm Version: (if applicable) ### To Reproduce 1. Point the exit-node hostname DNS record to **Address A** 2. Start `newt` — confirm holepunch targets **Address A** in the logs 3. Update the DNS record to point to **Address B** (simulating a failover) 4. Restart `newt` — confirm holepunch correctly targets **Address B** 5. Update the DNS record back to **Address A** 6. Wait longer than the DNS TTL (e.g. 8+ minutes with a TTL of ~60s) 7. Observe holepunch logs — it continues targeting **Address B** ### Expected Behavior After the DNS TTL expires, `newt`/holepunch should re-resolve the exit-node hostname and reconnect using the updated IP address — without requiring a manual restart.
GiteaMirror added the bug label 2026-05-13 18:37:11 -05:00
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
api
authentication
bug
config
dependencies
docker
documentation
enhancement
good first issue
help wanted
Improvement
Look Into
needs investigating
networking
new feature
non-critical bug
potential bug
pull-request
Mirrored from GitHub Pull Request
 question
reverse proxy
Security
stale
ui
wontfix
No Label bug
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
GiteaMirror ninjasurge
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: github-starred/pangolin#13072
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?