Docker

[toooo 17]

I have been seeing s6 errors on 4.9.5.0, as well. After a few hours of troubleshooting, found this thread. For me, the problem manifests as:

Running the container with podman, it does not ever stop when given `SIGTERM`, requiring podman to `SIGKILL` the container.

I backtracked to 4.9.3.0 and my setup works fine again.

# Running 4.9.3.0

[cont-finish.d] executing container finish scripts...
[cont-finish.d] done.
[s6-finish] waiting for services.
[s6-finish] sending all processes the TERM signal.
[s6-finish] sending all processes the KILL signal and exiting.

 

From Github copilot:

Quote

Findings:

Image 4.9.3.0 does NOT contain s6 runtime directories (/run/s6-rc/servicedirs and /run/service missing).
s6-related socket/file not present; s6-ipcclient failed (not present or not usable).
Container using 4.9.3.0 started and stopped cleanly (no SIGKILL observed).
Conclusion: the s6 shutdown race you saw is image-specific — newer image variants include s6 and exhibit the s6-ipcclient: Connection refused shutdown behavior, while 4.9.3.0 does not include s6 and avoids that problem.

 

[alucryd 334]

Are you running with `-it` or non interactively? If running with `-it` you should omit the `t`, it was never supported by s6, and outright stopped working in v3 [0]. Running without `t` will properly catch CTRL+C and stop the container. If running non-interactively, right now `podman stop` will ultimately resort to SIGKILL indeed, just fixed that behavior and the next image should stop immediately.

[0] https://github.com/just-containers/s6-overlay#terminal-support