Stale db locks can't be recovered

[Julian Edwards]

ed@beast:~/canonical/maas/sandbox$ bin/database destroy
/home/ed/canonical/maas/sandbox/db: cluster is locked by:
* 6487 (*unknown*)
ed@beast:~/canonical/maas/sandbox$ ps -fe|grep 6487
ed 15665 2864 0 16:51 pts/7 00:00:00 grep 6487
ed@beast:~/canonical/maas/sandbox$ bin/database stop
/home/ed/canonical/maas/sandbox/db: cluster is locked by:
* 6487 (*unknown*)

[Gavin Panella (allenap)]

As a workaround, when you see (*unknown*), delete the corresponding file in db/locks. For example, if you see:

  /home/ed/canonical/maas/sandbox/db: cluster is locked by:
  * 6487 (*unknown*)

do:

  $ rm /home/ed/canonical/maas/sandbox/db/locks/6487

The (*unknown*) bit means that it was unable to determine the command line for that process, which 99% of the time means that process is no longer running.