rauthy: Mounting UDS into Rauthy container causes the application to hang
Describe the bug
I’m running Rauthy using quadlet (systemd + Podman), with a 280 second timeout for shutdown. But it doesn’t shut down within that time. It feels like the SIGTERM is being ignored.
To Reproduce Steps to reproduce the behavior:
- Create
/etc/containers/systemd/rauthy.containerwith[Container] ContainerName=rauthy Image=ghcr.io/sebadob/rauthy:0.30.0 Network=none ReadOnly=true Volume=/srv/rauthy/config.toml:/app/config.toml:ro Volume=rauthy_data:/app/data:rw Volume=/run/rauthy.sock:/tmp/rauthy.sock:rw PodmanArgs=--stop-timeout 280 --restart on-failure [Service] TimeoutStopSec=300 systemctl daemon-reloadsystemctl start rauthy.servicesystemctl stop rauthy.service
Expected behavior Rauthy shuts down cleanly within 280 seconds.
I see systemd tries to stop things, but then nothing happens:
Jun 13 16:52:50 id systemd[1]: Stopping rauthy.service...
Until eventually, podman forcefully kills the container:
Jun 13 16:57:30 id rauthy[9905]: time="2025-06-13T16:57:30Z" level=warning msg="StopSignal SIGTERM failed to stop container rauthy in 280 seconds, resorting to SIGKILL"
Jun 13 16:57:30 id podman[9905]: 2025-06-13 16:57:30.82108436 +0000 UTC m=+280.063657441 container died 84ed0700429f308472d327185113cc666579f3af6ad3797b602dea0aca0aa508 (image=ghcr.io/sebadob/rauthy:0.30.0, name=rauthy, PODMAN_SYSTEMD_UNIT=rauthy.service)
Jun 13 16:57:30 id podman[9905]: 2025-06-13 16:57:30.875128 +0000 UTC m=+280.117701001 container remove 84ed0700429f308472d327185113cc666579f3af6ad3797b602dea0aca0aa508 (image=ghcr.io/sebadob/rauthy:0.30.0, name=rauthy, PODMAN_SYSTEMD_UNIT=rauthy.service)
Jun 13 16:57:30 id rauthy[9905]: 84ed0700429f308472d327185113cc666579f3af6ad3797b602dea0aca0aa508
Jun 13 16:57:30 id systemd[1]: rauthy.service: Main process exited, code=exited, status=137/n/a
Jun 13 16:57:30 id systemd[1]: rauthy.service: Failed with result 'exit-code'.
Additional context
This is a completely empty setup. It’s started up fresh with an empty database, in a single-node cluster.
- OS: Ubuntu 24.04 LTS
- Architecture: arm64
- systemd: 255.4-1ubuntu8.8
About this issue
- Original URL
- State: closed
- Created a year ago
- Comments: 22 (13 by maintainers)
Ah, damn it, you’re right. Rauthy needs to be the one creating the socket here. I need to do it the other way around 😄.
Yeah the network was just the first thing that came to my mind when thinking about the inital error. No network means no connections to anywhere else, but
localhostshould always work inside the container, so it would be enough for a single node.Yeah, I’ll do that in the mean time.
I was trying to avoid the network stack as a whole, hence initially going with
Network=none+ socket. Interestingly though it seems that even with Network=none there’s enough network in the container still for the Raft quorum to work. So if the UDS issue shakes out at some point I’ll try that again.