On Thu, 28.05.15 16:05, Daniel J Walsh ([email protected]) wrote: > When container stops machinectl still shows it registered? Do I need to > Unregister the machine? I though systemd would notice the pid died and > remove the machine.
Most likely it's not machined/machinectl that is confused here, but actually systemd itself. Do you still see the contaienr's scope in "systemctl"'s output when this happens? Unfortunately it's really nasty to properly shut down a scope in a robust way, since the cgroups empty notification is unreliable: it will not be generated if there's a subcrgoup still in a cgroup. Which means a container that creates subcgroups and then dies will never be noticed as dead. This will become robust as soon as we rework things for the unified hierarchy, since that for the first time supports reliable notifications and even inside of containers. In order to make this as robust as possible on current systems, please not only register your machines with machined from your container manager, but also unregister them explicitly, by invoking the "Terminate()" method on the machine object. This has the effect of asking systemd to recheck the scope, and then has the benefit of making it notice that the scope is dying... Lennart -- Lennart Poettering, Red Hat _______________________________________________ systemd-devel mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/systemd-devel
