Can't fork for 12: File exists about criu HOT 8 OPEN

The problem you see is because of duplicate PIDs. CRIU always creates processes with the same PID they had when checkpointing. In your destination environment the PID 12 is already in use so the restore fails.

Common workarounds are to use a container for each process you are trying to restore or the criu-ns script.

What are you trying to achieve? If we would know what your goal is we could maybe give you a better workaround.

Using criu-ns in a container will probably lead to nested namespaces and that is something CRIU usually does not handle well.

from criu.

@adrianreber Thank you for your response. I want to run multiple instances of the same app, with the second and subsequent instances coming from the checkpoint of the first instance. Each instance runs in a different docker container. In other words, a container runs only one instance.

from criu.

So why do you do the checkpoint inside of the container. Docker and Podman offer checkpoint commands which can easily do that:

$ podman container checkpoint -R --export=/tmp/cp.tar
$ podman container restore --import=/tmp/cp.tar --name=copy1
$ podman container restore --import=/tmp/cp.tar --name=copy2

Something like that.

from criu.

Common workarounds are to use a container for each process you are trying to restore or the criu-ns script.

It is currently restored in a different container. But report the error described above.

from criu.

Common workarounds are to use a container for each process you are trying to restore or the criu-ns script.

It is currently restored in a different container. But report the error described above.

Because there is already something running in the destination container. If you use Podman's or Docker's built-in support you avoid the problem of PID collisions.

from criu.

So why do you do the checkpoint inside of the container. Docker and Podman offer checkpoint commands which can easily do that:

$ podman container checkpoint -R --export=/tmp/cp.tar
$ podman container restore --import=/tmp/cp.tar --name=copy1
$ podman container restore --import=/tmp/cp.tar --name=copy2

Something like that.

Thanks, I'll try tomorrow.

from criu.

Common workarounds are to use a container for each process you are trying to restore or the criu-ns script.

It is currently restored in a different container. But report the error described above.

Because there is already something running in the destination container. If you use Podman's or Docker's built-in support you avoid the problem of PID collisions.

I understand a little bit. There are other programs occupying the pid.

from criu.

A friendly reminder that this issue had no activity for 30 days.

from criu.