fix: show Pid as 0 for stopped containers in `nerdctl container ins…

[haytok]

…pect`

In Docker, running docker container inspect on a stopped container shows the Pid as 0.

> docker ps -a --filter=name=nginx
CONTAINER ID   IMAGE     COMMAND                  CREATED          STATUS                     PORTS     NAMES
c8dcc84e427f   nginx     "/docker-entrypoint.…"   14 seconds ago   Exited (0) 8 seconds ago             nginx

> docker container inspect nginx | grep 'Pid"'
            "Pid": 0,

However, running nerdctl container inspect on a stopped container still shows a stale Pid from the already-exited process.

> sudo nerdctl ps -a --filter=name=nginx
CONTAINER ID    IMAGE                             COMMAND                   CREATED           STATUS                      PORTS    NAMES
4ce4cb9b1ed1    docker.io/library/nginx:latest    "/docker-entrypoint.…"    16 seconds ago    Exited (0) 4 seconds ago             nginx

> sn container inspect nginx | grep 'Pid"'
            "Pid": 1531853,

The docker daemon subscribes to containerd's events and when it detects a task exit event, it deletes the task and updates the state of the container it manages, setting Pid to 0. Therefore, in Docker the Pid becomes 0 after the container stops.

• https://github.com/moby/moby/blob/docker-v29.5.3/daemon/monitor.go#L34

On the other hand, nerdctl has no long-running daemon like the docker daemon, so it does not subscribe to containerd's events. As a result, the task is not deleted even after the container stops, and nerdctl container inspect still shows the stale Pid of the already-exited process returned by task.Pid().

Therefore, this commit changes the behavior so that nerdctl container inspect on a stopped container shows the Pid as 0, for compatibility with Docker.

[AkihiroSuda]

Thanks