After upgrading to latest newt some containers stopped working—it's totally random, and I have the same behavior on 4 different machines - LXC, VMs—mixed debian and ubuntu. I found out that it's docker socket - once I type the IP address, the asset starts to work. Once I revert back to newt 1.11.0 all works fine.
Environment
OS Type & Version: Debian bookowkrm VPS
Pangolin Version: 1.17.1
Gerbil Version: 1.3.1
Traefik Version: 3.6
Newt Version: 1.12.0
Olm Version: -
To Reproduce
Use 1.12.0 with docker socket—Grafana, Dockhand, Paperless-ngx, and many others stop working as long as I use docker socket, once I type serivce IP:port it's working again. All containers share the same docker network with newt and they can ping each other.
Expected Behavior
1.12.0 should work exactly same as 1.11.0 if no other changes were applied
Originally created by @cpuks on GitHub (Apr 28, 2026).
Original GitHub issue: https://github.com/fosrl/newt/issues/329
### Describe the Bug
After upgrading to latest newt some containers stopped working—it's totally random, and I have the same behavior on 4 different machines - LXC, VMs—mixed debian and ubuntu. I found out that it's docker socket - once I type the IP address, the asset starts to work. Once I revert back to newt 1.11.0 all works fine.
### Environment
- OS Type & Version: Debian bookowkrm VPS
- Pangolin Version: 1.17.1
- Gerbil Version: 1.3.1
- Traefik Version: 3.6
- Newt Version: 1.12.0
- Olm Version: -
### To Reproduce
Use 1.12.0 with docker socket—Grafana, Dockhand, Paperless-ngx, and many others stop working as long as I use docker socket, once I type serivce IP:port it's working again. All containers share the same docker network with newt and they can ping each other.
### Expected Behavior
1.12.0 should work exactly same as 1.11.0 if no other changes were applied
You have your newt setup properly to hit the socket you want?
You're going have to be more descriptive when you say you use docker socket as the data that's filled into pangolin and then traefik then relies squarely on the docker dns at 127.0.0.11:53 for resolving.. it doesn't poll the socket again and again for container ip changes it's purely just to find the dns entry that exists within docker aliases / dns.
<!-- gh-comment-id:4333326060 -->
@AstralDestiny commented on GitHub (Apr 28, 2026):
```yml
services:
newt:
image: fosrl/newt:1.12.0
container_name: newt
restart: unless-stopped
environment:
- PANGOLIN_ENDPOINT=https://
- NEWT_ID=
- NEWT_SECRET=
- LOG_LEVEL=DEBUG
- DOCKER_SOCKET=tcp://socket:2375
- docker-enforce-network-validation=true
# - DOCKER_SOCKET=/var/run/docker.sock
extra_hosts:
- host.docker.internal:host-gateway
networks:
- Janus
- Ares
- Artemis
volumes:
- /etc/localtime:/etc/localtime:ro
# - /var/run/docker.sock:/var/run/docker.sock:ro
```
You have your newt setup properly to hit the socket you want?
You're going have to be more descriptive when you say you use docker socket as the data that's filled into pangolin and then traefik then relies squarely on the docker dns at 127.0.0.11:53 for resolving.. it doesn't poll the socket again and again for container ip changes it's purely just to find the dns entry that exists within docker aliases / dns.
It works as intended on 1.11.0 and all previous versions; with 1.12.0 it's hit and miss—some services work, some don't. nothing is mentioned in release notes regarding extra entries in compose file
<!-- gh-comment-id:4333348988 -->
@cpuks commented on GitHub (Apr 28, 2026):
Sorry for lack of my compose:
```
services:
newt-vps:
image: fosrl/newt
container_name: newt-pangolin-vps
restart: unless-stopped
networks:
- suchanet
environment:
- PANGOLIN_ENDPOINT=<redacted>
- NEWT_ID=${NEWT_VPS_ID}
- NEWT_SECRET=${NEWT_VPS_SECRET}
- DOCKER_SOCKET=unix:///var/run/docker.sock
volumes:
- /var/run/docker.sock:/var/run/docker.sock:ro
```
It works as intended on 1.11.0 and all previous versions; with 1.12.0 it's hit and miss—some services work, some don't. nothing is mentioned in release notes regarding extra entries in compose file
@LaurenceJJones commented on GitHub (Apr 28, 2026):
Can you define LOG_LEVEL=DEBUG inside your compose environment for Newt and do a compose up. This will print debug information about the socket itself and can aid us in seeing what its doing.
<!-- gh-comment-id:4334482209 -->
@LaurenceJJones commented on GitHub (Apr 28, 2026):
Can you define `LOG_LEVEL=DEBUG` inside your compose environment for Newt and do a compose up. This will print debug information about the socket itself and can aid us in seeing what its doing.
@LaurenceJJones attached logs with debug turned on, same applies to my other machines running newt - random services are unavilable but only via docker socket - once I type ip it's ok
<!-- gh-comment-id:4335569353 -->
@cpuks commented on GitHub (Apr 28, 2026):
@LaurenceJJones attached logs with debug turned on, same applies to my other machines running newt - random services are unavilable but only via docker socket - once I type ip it's ok
[newt-pangolin-vps-logs.txt](https://github.com/user-attachments/files/27166543/newt-pangolin-vps-logs.txt)
@LaurenceJJones commented on GitHub (Apr 28, 2026):
Yeah so the service is failing due to the healthcheck configured is returning a 302 status as per #330 the healthcheck is not broken just users configured a healthcheck that ultimately had a redirect and now since we dont follow redirects it fails.
either update the healthcheck to go to a health endpoint or configure a healthcheck route such as /login or where grafana will ultimate redirect you too.
[newt-pangolin-vps] 2026-04-28T13:12:19.526946701Z DEBUG: 2026/04/28 13:12:19 Target 56: performing HTTP health check 3 to http://grafana:3000/
[newt-pangolin-vps] 2026-04-28T13:12:19.528416940Z WARN: 2026/04/28 13:12:19 Target 56: health check failed with status code 302
[newt-pangolin-vps] 2026-04-28T13:12:19.528530189Z DEBUG: 2026/04/28 13:12:19 Target 56: check failed (consecutive failures: 3 / threshold: 1): unhealthy status code: 302
<!-- gh-comment-id:4335729335 -->
@LaurenceJJones commented on GitHub (Apr 28, 2026):
Yeah so the service is failing due to the healthcheck configured is returning a `302` status as per #330 the healthcheck is not broken just users configured a healthcheck that ultimately had a redirect and now since we dont follow redirects it fails.
either update the healthcheck to go to a health endpoint or configure a healthcheck route such as `/login` or where grafana will ultimate redirect you too.
```
[newt-pangolin-vps] 2026-04-28T13:12:19.526946701Z DEBUG: 2026/04/28 13:12:19 Target 56: performing HTTP health check 3 to http://grafana:3000/
[newt-pangolin-vps] 2026-04-28T13:12:19.528416940Z WARN: 2026/04/28 13:12:19 Target 56: health check failed with status code 302
[newt-pangolin-vps] 2026-04-28T13:12:19.528530189Z DEBUG: 2026/04/28 13:12:19 Target 56: check failed (consecutive failures: 3 / threshold: 1): unhealthy status code: 302
```
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 @cpuks on GitHub (Apr 28, 2026).
Original GitHub issue: https://github.com/fosrl/newt/issues/329
Describe the Bug
After upgrading to latest newt some containers stopped working—it's totally random, and I have the same behavior on 4 different machines - LXC, VMs—mixed debian and ubuntu. I found out that it's docker socket - once I type the IP address, the asset starts to work. Once I revert back to newt 1.11.0 all works fine.
Environment
To Reproduce
Use 1.12.0 with docker socket—Grafana, Dockhand, Paperless-ngx, and many others stop working as long as I use docker socket, once I type serivce IP:port it's working again. All containers share the same docker network with newt and they can ping each other.
Expected Behavior
1.12.0 should work exactly same as 1.11.0 if no other changes were applied
@AstralDestiny commented on GitHub (Apr 28, 2026):
You have your newt setup properly to hit the socket you want?
You're going have to be more descriptive when you say you use docker socket as the data that's filled into pangolin and then traefik then relies squarely on the docker dns at 127.0.0.11:53 for resolving.. it doesn't poll the socket again and again for container ip changes it's purely just to find the dns entry that exists within docker aliases / dns.
@cpuks commented on GitHub (Apr 28, 2026):
Sorry for lack of my compose:
It works as intended on 1.11.0 and all previous versions; with 1.12.0 it's hit and miss—some services work, some don't. nothing is mentioned in release notes regarding extra entries in compose file
@LaurenceJJones commented on GitHub (Apr 28, 2026):
Can you define
LOG_LEVEL=DEBUGinside your compose environment for Newt and do a compose up. This will print debug information about the socket itself and can aid us in seeing what its doing.@cpuks commented on GitHub (Apr 28, 2026):
@LaurenceJJones attached logs with debug turned on, same applies to my other machines running newt - random services are unavilable but only via docker socket - once I type ip it's ok
newt-pangolin-vps-logs.txt
@LaurenceJJones commented on GitHub (Apr 28, 2026):
Yeah so the service is failing due to the healthcheck configured is returning a
302status as per #330 the healthcheck is not broken just users configured a healthcheck that ultimately had a redirect and now since we dont follow redirects it fails.either update the healthcheck to go to a health endpoint or configure a healthcheck route such as
/loginor where grafana will ultimate redirect you too.@cpuks commented on GitHub (Apr 28, 2026):
duplicate of #330