libpod api will not return stats if the container is exited

[ttys3]

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

/kind bug

Description

according to #14580

Non-running containers now report statistics via the podman stats

but I see the result for libpod api is : return nothing.

Steps to reproduce the issue:

1. podman run --name demo -ti alpine sh -c 'echo hello'

2. sudo podman stats --no-stream demo get:

ID            NAME        CPU %       MEM USAGE / LIMIT  MEM %       NET IO      BLOCK IO    PIDS        CPU TIME    AVG CPU %
9ef32ea2cb0e  demo        0.00%       0B / 0B            0.00%       0B / 0B     0B / 0B     0           0s          0.00%

3. but libpod api get nothing:

sudo curl -vvv --unix-socket /run/podman/podman.sock http://d/v1.0.0/libpod/containers/demo/stats\?stream\=false
*   Trying /run/podman/podman.sock:0...
* Connected to d (/run/podman/podman.sock) port 80 (#0)
> GET /v1.0.0/libpod/containers/demo/stats?stream=false HTTP/1.1
> Host: d
> User-Agent: curl/7.84.0
> Accept: */*
> 
* Mark bundle as not supporting multiuse
< HTTP/1.1 200 OK
< Api-Version: 1.41
< Libpod-Api-Version: 4.2.0-rc3
< Server: Libpod/4.2.0-rc3 (linux)
< X-Reference-Id: 0xc000014078
< Date: Fri, 05 Aug 2022 17:48:21 GMT
< Transfer-Encoding: chunked
< 
* Connection #0 to host d left intact

Describe the results you received:

nothing but a 200 code

Describe the results you expected:

zero stats just like the cli or it should return http error code ?

Additional information you deem important (e.g. issue happens only occasionally):

Output of podman version:

podman v4.1.1 and podman 4.2.0-rc3

Output of podman info:

(paste your output here)

Package info (e.g. output of rpm -q podman or apt list podman):

(paste your output here)

Have you tested with the latest version of Podman and have you checked the Podman Troubleshooting Guide? (https://github.com/containers/podman/blob/main/troubleshooting.md)

Yes

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

[rhatdan]

@mheon PTAL

[mheon]

We got a feature request in to allow this for the Docker-compat endpoints, so we enabled it there; never enabled it for the Libpod endpoints because nobody asked. Should be simple enough.

[mheon]

Actually, this might be expected behavior... Let me verify against the Docker endpoints.

[mheon]

Verified against Docker, this is expected behavior. If the container is not started, we return nothing, same as Docker. Going to close given this.

[ttys3]

the problem is, this change bring from cli, maybe related to #14580, I do not know.
it is a breaking change for libpod api.

previous version like podman v4.0.3, the cli will result in: Error: container state improper

and the libpod api will return:

< HTTP/1.1 409 Conflict
< Api-Version: 1.40
< Content-Type: application/json
< Libpod-Api-Version: 4.0.3
< Server: Libpod/4.0.3 (linux)
< X-Reference-Id: 0xc000518010
< Date: Sun, 07 Aug 2022 20:14:53 GMT
< Content-Length: 89
< 
{"cause":"container state improper","message":"container state improper","response":409}

which breaking exists code like:

https://github.com/hashicorp/nomad-driver-podman/blob/4efeb99d977c642a8aad85403f9b3f5d05256e1c/api/container_stats.go#L19-L30

which will cause nomad never got notified even the container exited with code 0

I checked the changelog v4.1.1:

Fixed a bug where the podman stats command could not be run on containers that were not running (it now reports all-0s statistics for Docker compatibility) (#14498).

the changelog says command, maybe more detailed (specific libpoad api and the compat api?).

Because this is a breaking change to the api.

[mheon]

It's a change, but I don't know if I would consider it a breaking change? It's a bugfix more than anything, given we did not match existing behavior / expected behavior vs Docker