Podman: Podman `play kube` errors out on podman-exported yaml

[kofalt]

Is this a BUG REPORT or FEATURE REQUEST? (leave only one on its own line)

/kind bug

Description

This tickets represents the import woes to #2963's export woes.
I'm not yet convinced they are related.

Steps to reproduce the issue:

1. Similar to my comment here https://github.com/containers/libpod/issues/2963#issuecomment-493722755, take the exported Kube yaml from the RH blog post.
   I have saved it as a gist for convenience.

2. Observe that you are unable to import the YAML from the blog:

podman play kube blog.yml

Error: pod was given but no pod is specified: invalid argument

I was having the same problem importing my own YAML, and so reproduced it with only blog material to make the case clearer.

Output of podman version:

Version:            1.3.1-dev
RemoteAPI Version:  1
Go Version:         go1.10.4
OS/Arch:            linux/amd64

Additional environment details (AWS, VirtualBox, physical, etc.):

Physical

[baude]

thanks for the well written defect, i appreciate the effort

[Fodoj]

Any update on this? Seems like play kube is unusable in 1.3.1

[mheon]

I think this is fixed upstream, and will be in 1.4.0 (which I'm releasing as we speak - https://github.com/containers/libpod/pull/3280

[Fodoj]

@mheon I tried with current master branch and it still doesn't work. New error is Error: error running container create option: log driver must be set: invalid argument

[haircommander]

hey @Fodoj can you PTAL at https://github.com/containers/libpod/pull/3283 and let me know if it fixes your play kube woes?

[mheon]

We need to improve testing here - we're breaking master repeatedly and not
catching it.

On Sat, Jun 8, 2019, 08:27 Peter Hunt [e-mail] wrote:

hey @Fodoj https://github.com/Fodoj can you PTAL at #3283
https://github.com/containers/libpod/pull/3283 and let me know if it
fixes your play kube woes?

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
https://github.com/containers/libpod/issues/3157?email_source=notifications&email_token=AB3AOCDHBPHLHBRSFD65CUTPZOQRLA5CNFSM4HN3RQV2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODXHTWRI#issuecomment-500120389,
or mute the thread
https://github.com/notifications/unsubscribe-auth/AB3AOCDRICFMEEHJ6X4XKVLPZOQRLANCNFSM4HN3RQVQ
.

[Fodoj]

Re testing: I can only agree whole play kube feature needs to be better tested. I have different bugs in 1.3.1, 1.3.2 and now 1.4.0 versions, feels like there are no tests for this feature at all!

@haircommander don't have time to test it right now, I rolled back to 1.3.0, otherwise development of my app was blocked.

[kofalt]

Could this be reopened?

I'm afraid that on pod 1.4.0 from apt, using the same instructions above, I get a different error:

$ podman play kube blog.yml
Trying to pull quay.io/baude/demodb:latest...Getting image source signatures
Copying blob a3ed95caeb02 done

# ...

Copying blob a068a53bcb17 done
Writing manifest to image destination
Storing signatures
Error: error running container create option: log driver must be set: invalid argument

This is reminiscent of #3283, and the patch author actually mentions play kube:

note: I considered fixing it on the play kube side, but instead opted for this

Let me know if anyone has trouble reproducing.

$ podman version
Version:            1.4.0
RemoteAPI Version:  1
Go Version:         go1.10.4
OS/Arch:            linux/amd64

[haircommander]

Hey @kofalt this was fixed with podman 1.14.1 can you try that out?

[mheon]

I think you mixed up your versioning there - it's 1.4.1 :smile:

[mheon]

(We also have improved upstream testing for play kube so failures like this should be much less frequent from here on)

[kofalt]

Sounds good. I can confirm 1.4.1 solves the issue!