Comment 12 for bug 1432387

Another possibility for cleaning up the lock files is to run a periodic task which will delete any lock files which correspond to volumes which have been deleted. We can have some delay between a volume being successfully deleted and when we remove it's lock files, giving time for other processes holding the same lock to work through the pipeline.

This is sort of a hack around the fact we can't tell if other processes are using a lock - we know that once a volume is deleted, nobody else will come into the pipe and grab a lock for that non-existent volume.

I'll see if I can get this cleanup working, otherwise will try out Josh's suggested approach using lock buckets.