[sys] podman play with init container: failed to remove once init container

[edsantiago]

I'm betting it's a bug in the test, not in podman. Filing so I don't forget:

not ok 548 podman play with init container
...
# podman kube play /tmp/podman_bats.lYSvJT/test.yaml
Error: failed to remove once init container CID:
      cannot remove container CID as it is running
        - running or paused containers cannot be removed without force:
           container state improper

Only one instance that I know of, just now in f38 aarch64 root sqlite

[Luap99]

maybe fixed by #18481?

[edsantiago]

Could be - I'll keep an eye on it. This one is failing hard, as in, every single run of #17831. Always aarch64.

• fedora-38-aarch64 : sys podman fedora-38-aarch64 root host sqlite
  • 05-04 22:16 in [sys] podman play with init container
• fedora-38-aarch64 : sys remote fedora-38-aarch64 root host sqlite [remote]
  • 05-05 07:13 in [sys] podman play with init container
  • 05-04 23:11 in [sys] podman play with init container

[edsantiago]

I cherry-picked an earlier version of #18481 into my hammer PR. Have not seen the "failed to remove once init container" error since, in about eight CI runs. I think that PR is promising, thanks.

[github-actions]

A friendly reminder that this issue had no activity for 30 days.

[edsantiago]

I see a lot of "failed to remove once init container" in my flake logs, but it seems to be only in logs where unlinkat-EBUSY has already triggered (i.e. where the entire system is hosed already). Closing, will reopen if it reoccurs without unlinkat-EBUSY.