Reik is right: There is a race-condition that happened to delete important data in the past breaking backups. I opted for the conservative approach to make the fix optional, as prohibiting concurrent execution is not a minor change. Still it is the most straighforward fix for the race-condition.
So I suggest an option "--allow-concurrency" (I hate negative boolean options ... think of "NOT NULL = false" .. just confusing) that would disable the lock (IMHO options should be named what they do, not how they do it).
Reik is right: There is a race-condition that happened to delete important data in the past breaking backups. I opted for the conservative approach to make the fix optional, as prohibiting concurrent execution is not a minor change. Still it is the most straighforward fix for the race-condition.
So I suggest an option "--allow- concurrency" (I hate negative boolean options ... think of "NOT NULL = false" .. just confusing) that would disable the lock (IMHO options should be named what they do, not how they do it).