Always purge stopped containers

[clinta]

Old containers should always be purged, even if they exited successfully. Because the logging socket is closed when the container exits, if the same container is restarted, instead of a new one being created it is unable to start because it cannot connect to the old logging socket.

[dadgar]

@clinta The if statement you changed is checking if the container is running or starting. Not whether it is dead. We purge all of the containers that have the same name if they are dead.

[clinta]

The check does not work as intended. A State.FinishedAt value of zero does not imply the container is starting. If a container fails to start, and is retried, State.FinishedAt will still be zero, but the container is not running or starting. As a result, any subsequent retry fails due to the container configuration pointing to an invalid logging socket.

A container inspect on a starting container will hang indefinitely, so there is no return value that represents a starting container. When a container is starting, that containers mutex is locked preventing the call to inspect from returning until it has a valid state of running or dead.

[github-actions]

I'm going to lock this pull request because it has been closed for 120 days ⏳. This helps our maintainers find and focus on the active contributions.
If you have found a problem that seems related to this change, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.