I have a docker container running at a local IP on the Pangolin VPS. I've used this docker container for months, but after updating to v1.17.0, the local site doesn't route traffic anymore. The error when trying to access the resource is "no available server" which I know from a failed health check. However, on the Local site there are no health checks I can configure. When I installed another newt container directly next to pangolin and switched to that site, the resource worked like it did in the months before. I assume something was changed on the local site health checks, because that would explain it not routing due to a failed health check.
Environment
OS Type & Version: Ubuntu Server 24.04.4 LTS
Pangolin, Gerbil, Traefik and Newt Version were updated to latest right after the 1.17.0 release
To Reproduce
Start a docker container on a local port (e.g. 172.17.0.1:3000) and add a resource through the standard Local site (without newt). When visiting the resource, it returns "no available server" even though the container is functional and can be pinged from inside the traefik container.
To workaround, install newt on the same machine where the container and pangolin are already running and change the site of the resource to this newt. Now it works!
Expected Behavior
Local resources should route traffic normally like they did previously. If health checks are involved, let the user edit/disable them, as they may cause errors.
Originally created by @yellowsnow3 on GitHub (Apr 4, 2026).
Original GitHub issue: https://github.com/fosrl/pangolin/issues/2782
### Describe the Bug
I have a docker container running at a local IP on the Pangolin VPS. I've used this docker container for months, but after updating to v1.17.0, the local site doesn't route traffic anymore. The error when trying to access the resource is "no available server" which I know from a failed health check. However, on the Local site there are no health checks I can configure. When I installed another newt container directly next to pangolin and switched to that site, the resource worked like it did in the months before. I assume something was changed on the local site health checks, because that would explain it not routing due to a failed health check.
### Environment
- OS Type & Version: Ubuntu Server 24.04.4 LTS
- Pangolin, Gerbil, Traefik and Newt Version were updated to latest right after the 1.17.0 release
### To Reproduce
Start a docker container on a local port (e.g. 172.17.0.1:3000) and add a resource through the standard Local site (without newt). When visiting the resource, it returns "no available server" even though the container is functional and can be pinged from inside the traefik container.
To workaround, install newt on the same machine where the container and pangolin are already running and change the site of the resource to this newt. Now it works!
### Expected Behavior
Local resources should route traffic normally like they did previously. If health checks are involved, let the user edit/disable them, as they may cause errors.
I am seeing the same issue where local sites have worked since a year ago but are now all offline. I've been trying to debug for the last half hour but haven't figured out a cause. If I manually add the router/service to the Traefik dynamic config file I can access the resource just fine but then it's skipping Pangolin auth of course... but at least that establishes that Traefik doesn't have any problem accessing local resources (using Docker container_name:port) except when going through Pangolin.
<!-- gh-comment-id:4185991898 -->
@dlhall111 commented on GitHub (Apr 4, 2026):
I am seeing the same issue where local sites have worked since a year ago but are now all offline. I've been trying to debug for the last half hour but haven't figured out a cause. If I manually add the router/service to the Traefik dynamic config file I can access the resource just fine but then it's skipping Pangolin auth of course... but at least that establishes that Traefik doesn't have any problem accessing local resources (using Docker container_name:port) except when going through Pangolin.
@dlhall111 you can create a "local" newt site and change the target site from local to that for the time being, then it doesn't bypass pangolin but works until this is fixed
<!-- gh-comment-id:4186022894 -->
@yellowsnow3 commented on GitHub (Apr 4, 2026):
@dlhall111 you can create a "local" newt site and change the target site from local to that for the time being, then it doesn't bypass pangolin but works until this is fixed
@oschwartz10612 I'm running the Community Edition (guess that'd be OSS)...
@yellowsnow3 and yeah I hadn't thought of doing that until you mentioned it, great idea and works like a charm! :)
<!-- gh-comment-id:4186036644 -->
@dlhall111 commented on GitHub (Apr 4, 2026):
@oschwartz10612 I'm running the Community Edition (guess that'd be OSS)...
@yellowsnow3 and yeah I hadn't thought of doing that until you mentioned it, great idea and works like a charm! :)
@oschwartz10612 commented on GitHub (Apr 4, 2026):
@dlhall111 Thanks! Do you have any other resources with health checks?
<!-- gh-comment-id:4186040762 -->
@oschwartz10612 commented on GitHub (Apr 4, 2026):
@dlhall111 Thanks! Do you have any other resources with health checks?
@oschwartz10612 I don’t use healthchecks with any resources (local or otherwise), never have.
<!-- gh-comment-id:4186047555 -->
@dlhall111 commented on GitHub (Apr 4, 2026):
@oschwartz10612 I don’t use healthchecks with any resources (local or otherwise), never have.
@oschwartz10612 commented on GitHub (Apr 4, 2026):
Also, would you be able to do the following for me?
$ sudo apt install sqlite3
$ sqlite3 config/db/db.sqlite
> select online, siteId, name from sites where type = 'local';
<!-- gh-comment-id:4186048253 -->
@oschwartz10612 commented on GitHub (Apr 4, 2026):
Also, would you be able to do the following for me?
```
$ sudo apt install sqlite3
$ sqlite3 config/db/db.sqlite
> select online, siteId, name from sites where type = 'local';
```
I am on OSS. I have multiple other resources with health checks which work. The SQL Query returned 0|1|Local
<!-- gh-comment-id:4186052401 -->
@yellowsnow3 commented on GitHub (Apr 4, 2026):
I am on OSS. I have multiple other resources with health checks which work. The SQL Query returned `0|1|Local`
@oschwartz10612 if it helps my return is essentially the same (0|7|local).
<!-- gh-comment-id:4186059423 -->
@dlhall111 commented on GitHub (Apr 4, 2026):
@oschwartz10612 if it helps my return is essentially the same (0|7|local).
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 @yellowsnow3 on GitHub (Apr 4, 2026).
Original GitHub issue: https://github.com/fosrl/pangolin/issues/2782
Describe the Bug
I have a docker container running at a local IP on the Pangolin VPS. I've used this docker container for months, but after updating to v1.17.0, the local site doesn't route traffic anymore. The error when trying to access the resource is "no available server" which I know from a failed health check. However, on the Local site there are no health checks I can configure. When I installed another newt container directly next to pangolin and switched to that site, the resource worked like it did in the months before. I assume something was changed on the local site health checks, because that would explain it not routing due to a failed health check.
Environment
To Reproduce
Start a docker container on a local port (e.g. 172.17.0.1:3000) and add a resource through the standard Local site (without newt). When visiting the resource, it returns "no available server" even though the container is functional and can be pinged from inside the traefik container.
To workaround, install newt on the same machine where the container and pangolin are already running and change the site of the resource to this newt. Now it works!
Expected Behavior
Local resources should route traffic normally like they did previously. If health checks are involved, let the user edit/disable them, as they may cause errors.
@dlhall111 commented on GitHub (Apr 4, 2026):
I am seeing the same issue where local sites have worked since a year ago but are now all offline. I've been trying to debug for the last half hour but haven't figured out a cause. If I manually add the router/service to the Traefik dynamic config file I can access the resource just fine but then it's skipping Pangolin auth of course... but at least that establishes that Traefik doesn't have any problem accessing local resources (using Docker container_name:port) except when going through Pangolin.
@yellowsnow3 commented on GitHub (Apr 4, 2026):
@dlhall111 you can create a "local" newt site and change the target site from local to that for the time being, then it doesn't bypass pangolin but works until this is fixed
@oschwartz10612 commented on GitHub (Apr 4, 2026):
Looking into it...
@oschwartz10612 commented on GitHub (Apr 4, 2026):
Are you on EE or OSS?
@dlhall111 commented on GitHub (Apr 4, 2026):
@oschwartz10612 I'm running the Community Edition (guess that'd be OSS)...
@yellowsnow3 and yeah I hadn't thought of doing that until you mentioned it, great idea and works like a charm! :)
@oschwartz10612 commented on GitHub (Apr 4, 2026):
@dlhall111 Thanks! Do you have any other resources with health checks?
@dlhall111 commented on GitHub (Apr 4, 2026):
@oschwartz10612 I don’t use healthchecks with any resources (local or otherwise), never have.
@oschwartz10612 commented on GitHub (Apr 4, 2026):
Also, would you be able to do the following for me?
@yellowsnow3 commented on GitHub (Apr 4, 2026):
I am on OSS. I have multiple other resources with health checks which work. The SQL Query returned
0|1|Local@oschwartz10612 commented on GitHub (Apr 4, 2026):
Thanks @yellowsnow3 thats helpful. I think reverting #2570 will fix this. Pushing over the release now.
@dlhall111 commented on GitHub (Apr 4, 2026):
@oschwartz10612 if it helps my return is essentially the same (0|7|local).