Comment # 9 on bug 1012883 from Franck Bui

With the testing package, I managed to reproduce the issue one more time and
the journal had:

Dec 07 18:59:39 linux-bzg2 systemd[1]: session-1.scope: Stopping timed out.
Killing.
Dec 07 18:59:39 linux-bzg2 systemd[1]: session-1.scope: ksmserver ignored TERM
signal, killed.
Dec 07 18:59:39 linux-bzg2 systemd[1]: session-1.scope changed stop-sigterm ->
stop-sigkill
Dec 07 18:59:39 linux-bzg2 systemd[1]: Received SIGCHLD from PID 1883
(ksmserver).
Dec 07 18:59:39 linux-bzg2 systemd[1]: Child 1883 (ksmserver) died
(code=killed, status=9/KILL)
Dec 07 18:59:39 linux-bzg2 systemd[1]: session-1.scope: Child 1883 belongs to
session-1.scope
Dec 07 18:59:39 linux-bzg2 systemd[1]: session-1.scope: cgroup is empty
Dec 07 18:59:39 linux-bzg2 systemd[1]: session-1.scope changed stop-sigkill ->
failed
Dec 07 18:59:39 linux-bzg2 systemd[1]: session-1.scope: Job
session-1.scope/stop finished, result=done
Dec 07 18:59:39 linux-bzg2 systemd[1]: Stopped Session 1 of user fbui.
Dec 07 18:59:39 linux-bzg2 systemd[1]: session-1.scope: Unit entered failed
state.

This confirms that ksmserver is the hanging process that seems to fail to
handle the TERM signal properly in my case.