I'm having a similar issue to #201, though this doesn't appear to be related to the IP address. I was using newt with the native client mode and after upgrading to 1.7.0 my Newt instances crash when trying to create the network interface. A fresh redeploy doesn't seem to help either.
I added the /dev/net/tun devices specification to see if that would help but no luck.
Environment
OS Type & Version: Docker
Newt Version: 1.7.0
To Reproduce
Run Newt in native client mode with the parameters shown in the compose snippet.
Expected Behavior
Newt creates the WG interface successfully and doesn't crash.
Originally created by @tullisar on GitHub (Dec 17, 2025).
Original GitHub issue: https://github.com/fosrl/newt/issues/210
### Describe the Bug
I'm having a similar issue to #201, though this doesn't appear to be related to the IP address. I was using newt with the native client mode and after upgrading to 1.7.0 my Newt instances crash when trying to create the network interface. A fresh redeploy doesn't seem to help either.
Here's a snippet of the log:
```
client-native-1 | ERROR: 2025/12/16 17:27:22 Failed to ensure WireGuard interface: failed to create native TUN device: CreateTUN("newt") failed; /dev/net/tun does not exist
client-native-1 | panic: runtime error: invalid memory address or nil pointer dereference
client-native-1 | [signal SIGSEGV: segmentation violation code=0x1 addr=0x260 pc=0x632b2c]
client-native-1 |
client-native-1 | goroutine 69 [running]:
client-native-1 | sync/atomic.(*Int32).Add(...)
client-native-1 | /usr/local/go/src/sync/atomic/type.go:94
client-native-1 | sync.(*RWMutex).RLock(...)
client-native-1 | /usr/local/go/src/sync/rwmutex.go:72
client-native-1 | golang.zx2c4.com/wireguard/device.(*Device).IpcGetOperation(0x0, {0x11a3ca0, 0xc000052440})
client-native-1 | /go/pkg/mod/golang.zx2c4.com/wireguard@v0.0.0-20250521234502-f333402bd9cb/device/uapi.go:52 +0x4c
client-native-1 | golang.zx2c4.com/wireguard/device.(*Device).IpcGet(0x0)
client-native-1 | /go/pkg/mod/golang.zx2c4.com/wireguard@v0.0.0-20250521234502-f333402bd9cb/device/uapi.go:409 +0x38
client-native-1 | github.com/fosrl/newt/clients.(*WireGuardService).ensureWireguardPeers(0xc00033e9c0, {0xc0003d81c0, 0x3, 0xc0003d81c0?})
client-native-1 | /app/clients/clients.go:648 +0x45
client-native-1 | github.com/fosrl/newt/clients.(*WireGuardService).handleConfig(0xc00033e9c0, {{0xc0000400a8, 0x16}, {0xec0520, 0xc00040af00}})
client-native-1 | /app/clients/clients.go:475 +0x274
client-native-1 | github.com/fosrl/newt/websocket.(*Client).readPumpWithDisconnectDetection(0xc000274dc0, {0xc000066fd0?, 0x577324?, 0x1976320?})
client-native-1 | /app/websocket/client.go:742 +0x21c
client-native-1 | created by github.com/fosrl/newt/websocket.(*Client).establishConnection in goroutine 9
client-native-1 | /app/websocket/client.go:565 +0xb6a
```
For reference, he's a snippet of the deployed compose configuration for the service:
```
name: newt
services:
client-native:
profiles:
- base
- native
cap_add:
- NET_ADMIN
- SYS_MODULE
depends_on:
socket-proxy:
condition: service_healthy
required: true
devices:
- source: /dev/net/tun
target: /dev/net/tun
permissions: rwm
environment:
ACCEPT_CLIENTS: "true"
DOCKER_SOCKET: unix:///socket/newt.sock
HEALTH_FILE: /tmp/healthy
KEEP_INTERFACE: "true"
LOG_LEVEL: INFO
NEWT_ID: <REDACTED>
NEWT_SECRET: <REDACTED>
PANGOLIN_ENDPOINT: <REDACTED>
USE_NATIVE_INTERFACE: "true"
healthcheck:
test:
- CMD-SHELL
- cat /tmp/healthy || exit 1
timeout: 10s
interval: 1m0s
retries: 3
start_period: 1m0s
image: fosrl/newt:latest
network_mode: host
restart: unless-stopped
```
I added the /dev/net/tun devices specification to see if that would help but no luck.
### Environment
- OS Type & Version: Docker
- Newt Version: 1.7.0
### To Reproduce
Run Newt in native client mode with the parameters shown in the compose snippet.
### Expected Behavior
Newt creates the WG interface successfully and doesn't crash.
Deleting the newt interface on the host with sudo ip link del newt, then redeploying the container fixed the issue. Seems like a migration issue with regards to the KEEP_INTERFACE parameter? I'm going to leave my containers running for a day or so then I'll close the issue if nothing breaks.
<!-- gh-comment-id:3663244112 -->
@tullisar commented on GitHub (Dec 17, 2025):
Deleting the `newt` interface on the host with `sudo ip link del newt`, then redeploying the container fixed the issue. Seems like a migration issue with regards to the `KEEP_INTERFACE` parameter? I'm going to leave my containers running for a day or so then I'll close the issue if nothing breaks.
@oschwartz10612 commented on GitHub (Dec 18, 2025):
Yeah KEEP_INTERFACE was deprecated as it does not use netlink anymore to create interfaces. I can see if you were migrating the need to delete the old interface.
<!-- gh-comment-id:3670866344 -->
@oschwartz10612 commented on GitHub (Dec 18, 2025):
Yeah `KEEP_INTERFACE` was deprecated as it does not use netlink anymore to create interfaces. I can see if you were migrating the need to delete the old interface.
In 3305f71 I tried to prevent the segfault.
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Originally created by @tullisar on GitHub (Dec 17, 2025).
Original GitHub issue: https://github.com/fosrl/newt/issues/210
Describe the Bug
I'm having a similar issue to #201, though this doesn't appear to be related to the IP address. I was using newt with the native client mode and after upgrading to 1.7.0 my Newt instances crash when trying to create the network interface. A fresh redeploy doesn't seem to help either.
Here's a snippet of the log:
For reference, he's a snippet of the deployed compose configuration for the service:
I added the /dev/net/tun devices specification to see if that would help but no luck.
Environment
To Reproduce
Run Newt in native client mode with the parameters shown in the compose snippet.
Expected Behavior
Newt creates the WG interface successfully and doesn't crash.
@tullisar commented on GitHub (Dec 17, 2025):
Deleting the
newtinterface on the host withsudo ip link del newt, then redeploying the container fixed the issue. Seems like a migration issue with regards to theKEEP_INTERFACEparameter? I'm going to leave my containers running for a day or so then I'll close the issue if nothing breaks.@oschwartz10612 commented on GitHub (Dec 18, 2025):
Yeah
KEEP_INTERFACEwas deprecated as it does not use netlink anymore to create interfaces. I can see if you were migrating the need to delete the old interface.In
3305f71I tried to prevent the segfault.