restart: failed to allocate for range 0: 10.88.109.255 has been allocated to XXX, duplicate allocation is not allowed

[edsantiago]

  [It] podman start container sets HOME to home of execUser
...
           # podman-remote [options] create --userns keep-id --user bin:bin quay.io/libpod/alpine:latest ls
           b56f8c668fc53da24557b1af33319fe027ca59860d30fe1b9ecc7fe2fc562674
           # podman-remote [options] start b56f8c668fc53da24557b1af33319fe027ca59860d30fe1b9ecc7fe2fc562674
           b56f8c668fc53da24557b1af33319fe027ca59860d30fe1b9ecc7fe2fc562674
           # podman-remote [options] inspect b56f8c668fc53da24557b1af33319fe027ca59860d30fe1b9ecc7fe2fc562674 --format {{.Config.Env}}
           [PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin TERM=xterm container=podman HOME=/bin HOSTNAME=b56f8c668fc5]
           # podman-remote [options] restart b56f8c668fc53da24557b1af33319fe027ca59860d30fe1b9ecc7fe2fc562674
           time="2023-05-26T21:38:19-05:00" level=warning msg="Failed to load cached network config: network podman not found in CNI cache, falling back to loading network podman from disk"
           Error: plugin type="bridge" failed (add): cni plugin bridge failed: failed to allocate for range 0: 10.88.109.255 has been allocated to b56f8c668fc53da24557b1af33319fe027ca59860d30fe1b9ecc7fe2fc562674, duplicate allocation is not allowed

Seen in f37 root remote

[Luap99]

The warning can be ignored, the correct error is failed to allocate for range 0: 10.88.109.255 has been allocated to XXX, duplicate allocation is not allowed. The issue sounds like the network teardown was not done for the stop and thus start failed.

[edsantiago]

Removed remote from tags and title. This is no longer a remote-only bug.

• fedora-37 : int podman fedora-37 root container sqlite
  • 06-06 19:13 in podman start container does not set HOME to home of caller
• fedora-37 : int remote fedora-37 root host sqlite [remote]
  • 05-26 23:00 in podman start container sets HOME to home of execUser
  • 05-18 19:35 in podman start container does not set HOME to home of caller

[github-actions]

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

[Luap99]

@edsantiago still happening?

[edsantiago]

Apparently, yes. Here's one from October 4 (not a colorized log, because the CI run timed out) and one from August 29.

The error is infrequent, and it does not trigger a CI failure because in both places (podman run/create two containers with the same IP) the ginkgo code just does Expect(ExitWithError) and a substring check for message but no check for other messages. Since it does not trigger a CI failure, I'm only seeing it in logs where there's another flake.

[edsantiago]

However, I do not see the message in any "set HOME" tests as originally reported.

[Luap99]

Ok given and is that infrequent and CNI + remote only?? I rather not debug it given we will remove CNI with 5.0.

[edsantiago]

I'm OK with that.