Fix -f logs follow with stopped container

[QiWang19]

Fix -f logs follow with stopped container. Close #6531

Signed-off-by: Qi Wang qiwan@redhat.com

[openshift-ci-robot]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by: QiWang19
To complete the pull request process, please assign baude
You can assign the PR to them by writing /assign @baude in a comment when ready.

The full list of commands accepted by this bot can be found here.

Details

Needs approval from an approver in each of these files:

• OWNERS

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[mheon]

Does this fix journald logging as well, or just file logging?

[rhatdan]

Does this work correct.

In one window run

podman run --name test -ti fedora bash -c "sleep 10; ls -R /"

In a separatee window

podman log test

When the first container ends, does the podman log command also end?

[QiWang19]

This just unset the follow options for the file logging.

[rhatdan]

Right but it has no effect on a running log.

[QiWang19]

Looks not right, If run logs -f test in a separate window, the log command directly ends without logging anything.

[jgallucci32]

@QiWang19 When following logs on a running container, if the container is stopped, does the logging stop as well? I want to make sure this covers both use cases

• When following logs on stopped container, exit immediately
• When following logs on a running container, exit when container is stopped

[QiWang19]

* When following logs on a running container, exit when container is stopped
* When following logs on stopped container, exit immediately

Yes, the current upstream has both these issues that should be fixed and I am working on.

[rhatdan]

@QiWang19 Could you rebase and repush. Tests do not seem to have run. Have you confirmed that this is running the way it should?

[QiWang19]

This PR not fully fixed the issue. Still need more work on this.

[jgallucci32]

@QiWang19 I just tested this and following the logs hangs indefinitely if the container is running before running logs -f container and after it stops. It will no longer hang if the container is already stopped, but stopping a container while tailing the logs continues to be an issue.

[jgallucci32]

@QiWang19 I have traced the hang in the code to this line
https://github.com/containers/libpod/blob/033743cbee180d9a4149f574375cc1201e573d06/libpod/container_log.go#L61
Since the Lines() function is external to libpod we're likely going to have to loop and check for it like the original fix was written 2 years ago (it had a time.Sleep(1 * time.Second) line within a loop to check the state of the container every second)

[jgallucci32]

@QiWang19 I created a PR against your branch which handles multiple scenarios of following a log with stopped (or stopping) containers.
QiWang19#1

What this does is create a subroutine which checks the state of the container every second. This is a similar approach taken when this bug was fixed 2 years ago. The difference with the new code from 2 years ago is libpod is using github.com/hpcloud/tail which effectively follows a log until it is given an exit command. Before the logic for tailing a file was written as functions within libpod. hpcloud/tail has several options Kill, Stop, StopAtEOF so I chose what I thought would allow the function to exit most gracefully which is StopAtEOF().

I tested the following conditions with this PR:

• Tail and follow logs of running container
• Tail and follow logs of stopped container
• Tail and follow logs of running container which exits after some time

[rhatdan]

@jgallucci32 Could you open a PR for libpod with Yours and @QiWang19 Changes combined and here tests, Leave her signature and then we can merge the combination.

[jgallucci32]

@rhatdan I created PR #6591 which is a merge of our two commits. It will supercede this PR.

Not sure how to leave her signature, but there are 2 commits in the PR one from each of us.