gaia-unmount might not be executed in case of "fleetctl destroy"

[shayts7]

There are cases in which "fleetctl destroy" won't call the ExecStop or ExecPostStop commands, hence - won't call the "gaia-unmount" part.
In such cases, if the unit will try to restart on another host - it will fail as the volume is still mounted on the original host.

Here is journalctl output of influxdb.service being destroy by ansible, and we can see there is not entries for "gaia-unmount.sh":

Jan 24 12:46:16 ip-10-10-47-127.us-west-2.compute.internal bash[10813]: [wal] 2016/01/24 12:46:16 write to index of partition 1 took 285.795392ms
Jan 24 12:46:26 ip-10-10-47-127.us-west-2.compute.internal bash[10813]: [wal] 2016/01/24 12:46:26 Flush due to idle. Flushing 832 series with 832 points and 24917 bytes from partition 1
Jan 24 12:46:26 ip-10-10-47-127.us-west-2.compute.internal bash[10813]: [wal] 2016/01/24 12:46:26 write to index of partition 1 took 359.71396ms
Jan 24 12:46:28 ip-10-10-47-127.us-west-2.compute.internal systemd[1]: Stopping InfluxDB v.0.9...
Jan 24 12:46:36 ip-10-10-47-127.us-west-2.compute.internal bash[10813]: [wal] 2016/01/24 12:46:36 Flush due to idle. Flushing 832 series with 832 points and 24917 bytes from partition 1
Jan 24 12:46:36 ip-10-10-47-127.us-west-2.compute.internal bash[10813]: [wal] 2016/01/24 12:46:36 write to index of partition 1 took 285.508134ms
Jan 24 12:46:38 ip-10-10-47-127.us-west-2.compute.internal docker[1586]: influxdb
Jan 24 12:46:38 ip-10-10-47-127.us-west-2.compute.internal systemd[1]: influxdb.service: Main process exited, code=exited, status=137/n/a
Jan 24 12:46:38 ip-10-10-47-127.us-west-2.compute.internal systemd[1]: Stopped influxdb.service.
Jan 24 12:46:38 ip-10-10-47-127.us-west-2.compute.internal systemd[1]: influxdb.service: Unit entered failed state.
Jan 24 12:46:38 ip-10-10-47-127.us-west-2.compute.internal systemd[1]: influxdb.service: Failed with result 'exit-code'.

There is an open bug in fleet talking exactly about the same issue: coreos/fleet#1000