`podman play kube` - Need Container Name as well as ID when Container Fails to Start

[stellarpower]

Issue Description

I have seen problems like this a few times:

$ podman play kube --replace 'Some Project k8s.yaml'
Pods stopped:
Pods removed:
Secrets removed:
Volumes removed:
Pod:
6af8a146793983ef3ae16c36d606337d90c88a446643b59a4fe1dad060b5da51
Containers:
788af1bb4db51250cbd14c05179db989810ded1613a6f7725f8bb768cd3d48e5
c15e7890303a4806047b1b68585f6d80c9182867a27f8b0986fef74f8e052423

starting container bbb8d5fcc9584f5c2ce594bd368d564d76b8791008dbdea6217a2b9cd95f0ea2: runc: runc create failed: invalid rootfs: not an absolute path, or a symlink: OCI runtime error
starting container 788af1bb4db51250cbd14c05179db989810ded1613a6f7725f8bb768cd3d48e5: a dependency of container 788af1bb4db51250cbd14c05179db989810ded1613a6f7725f8bb768cd3d48e5 failed to start: container state improper
starting container c15e7890303a4806047b1b68585f6d80c9182867a27f8b0986fef74f8e052423: a dependency of container c15e7890303a4806047b1b68585f6d80c9182867a27f8b0986fef74f8e052423 failed to start: container state improper

Error: failed to start 3 containers

Here, I'm not concerned about what is causing the issue with starting the pod. However, only the container ID is printed, which obviously changes each time I redeploy. If the container can't be created, I can't look up the ID with podman ps - making it difficult to try and narrow down what the issue is - I have to start commenting and uncommenting lines in the YAML. I'd like ot see the container name from the k8s YAML printed also:

$ podman play kube --replace 'Some Project k8s.yaml'
Pods stopped:
Pods removed:
Secrets removed:
Volumes removed:
Pod:
hello-world-pod           (6af8a146793983ef3ae16c36d606337d90c88a446643b59a4fe1dad060b5da51)
Containers:
hello-world-pod-postgres  (788af1bb4db51250cbd14c05179db989810ded1613a6f7725f8bb768cd3d48e5)
hello-world-pod-redis     (c15e7890303a4806047b1b68585f6d80c9182867a27f8b0986fef74f8e052423)

starting container hello-world-pod-server (bbb8d5fcc9584f5c2ce594bd368d564d76b8791008dbdea6217a2b9cd95f0ea2): runc: runc create failed: invalid rootfs: not an absolute path, or a symlink: OCI runtime error
starting container hello-world-pod-postgres (788af1bb4db51250cbd14c05179db989810ded1613a6f7725f8bb768cd3d48e5): a dependency of container hello-world-pod-postgres (788af1bb4db51250cbd14c05179db989810ded1613a6f7725f8bb768cd3d48e5) failed to start: container state improper
starting container hello-world-pod-redis (c15e7890303a4806047b1b68585f6d80c9182867a27f8b0986fef74f8e052423): a dependency of container hello-world-pod-redis (c15e7890303a4806047b1b68585f6d80c9182867a27f8b0986fef74f8e052423) failed to start: container state improper

Error: failed to start 3 containers

Thanks!

Steps to reproduce the issue

See above

Describe the results you received

See above

Describe the results you expected

See above

podman info output

podman version
Client:       Podman Engine
Version:      5.5.2
API Version:  5.5.2
Go Version:   go1.24.4
Built:        Wed Jun 25 05:50:07 2025
OS/Arch:      linux/amd64

Podman in a container

No

Privileged Or Rootless

Rootless

Upstream Latest Release

No

Additional environment details

No response

Additional information

No response

[Luap99]

Most of the errors are generated quite far internally in the libpod code so if we do this this would effect basically all commands. But I don't think that should be a problem, I tend to agree that printing both the name and ID as part of error messages makes sense as a user could need either one to debug/understand the problem. And given it is just for errors the extra characters should not be an issue

[baude]

any chance we could come up with a simpler reproducer. can we do this with plain old start or run ?

[stellarpower]

Most of the errors are generated quite far internally in the libpod code so if we do this this would effect basically all commands.

Yeah, did figure that might be the case. Is libpod just logging/"format-stringing" itself here? I.e. it isn't returning a message with the container ID alongside back to higher scope. I see advantages of having libpod format the string but at the same time it can be nice to return errors in a structured format so the tool that use it can decide what it wants to do.

[Luap99]

Is libpod just logging/"format-stringing" itself here?

Yes go errors are mainly just strings wrapping, there are ways to do it differently of course if really needed though it ends up being more complex.

In general many errors (or normal log messages) include only either the ID or name, it might be nice to come up with a policy on how to log them consistently across the code base. I think there is value logging both ID and name, maybe trim the ID down to 12 chars to not make the log lines to long and keep the human readable?

[stellarpower]

I've not had much reason to use go beyond the basics in the past, is it presumably a bit complicating things to wrap a struct with a lazy function that can format either way at the outer scope's request?

[Luap99]

I've not had much reason to use go beyond the basics in the past, is it presumably a bit complicating things to wrap a struct with a lazy function that can format either way at the outer scope's request?

That would make things more complicated and I don't see what we would gain. I don't think this problem is special to play kube at all, at least when it comes to log lines and error messages. If you look at podman pod start, podman start --all, podman system service they all can act on many containers and it depends on the user if the ID or name provides more purpose to them although most often the later I guess.

So really I would view this as including the output most users need in log messages and errors. And because it is trivial to both have the name and ID shown I would juts do that instead of complicating things to much.

[github-actions]

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