Bug#788269: machinectl loses track of running systemd-nspawn machines

[Joey Hess]

Package: systemd
Version: 220-5
Severity: normal

For some reason, after a period of hours or days, systemd-nspawn
containers that are still running are no longer listed by machinectl.

root at elephant:~>systemctl status systemd-nspawn at oldusenet-shellbox.service
● systemd-nspawn at oldusenet-shellbox.service - Container oldusenet/shellbox
   Loaded: loaded (/etc/systemd/system/multi-user.target.wants/systemd-nspawn at oldusenet-shellbox.service)
   Active: active (running) since Fri 2015-06-05 11:17:02 JEST; 4 days ago
     Docs: man:systemd-nspawn(1)
 Main PID: 25613 (systemd-nspawn)
   Status: "Container running.
   CGroup: /machine.slice/systemd-nspawn at oldusenet-shellbox.service
           ├─25613 /usr/bin/systemd-nspawn --quiet --keep-unit --boot --direc...
           ├─25660 /lib/systemd/systemd
           └─system.slice
             ├─dbus.service
             │ └─25739 /usr/bin/dbus-daemon --system --address=systemd: --nof...
             ├─cron.service
             │ └─25731 /usr/sbin/cron -f
             ├─oldusenet.service
              ...

root at elephant:~>machinectl
MACHINE CLASS SERVICE

0 machines listed.

root at elephant:~>journalctl --since '2015-06-05' | grep systemd-machined
Jun 05 11:17:02 elephant systemd-machined[25661]: New machine oldusenet-shellbox.
Jun 08 03:40:19 elephant systemd-machined[25661]: Machine oldusenet-shellbox terminated.

So, machinectl thinks it terminated, but the service is actually still running.

If I stop and restart this service, machinectl will list it for a while, and
then again lose it. Observed this repeatedly on 2 boxes, both running 220-5.
A third box with 215-18 and otherwise similar configuration does not have the
problem.

-- 
see shy jo