A regression introduced by , can cause a file detach to timeout, if another client is momentarily updating empty scopes or relationships.
changed the way the delete timeouts were handled, so that it it would not block others, by instead checking for a lock then sleeping.
However, detach also makes other locking calls to check for empty scopes and update relationships and the same 0 timeout was being used as a result of the change.
If another client was in that, short-lived locking phase, the detach would fire an instant DFS timeout, which was returned to the client.

The 0 time timeout was supposed to be for the file locks, to avoid deadlock, it was not supposed to effect the exclusive locking that empty scope and relationship checking performs.

Change those timeouts to use the DFS default timeout.