When replacing the actual sharding catalog manager in a mongo process, the old manager will shut down without being able to release any distributed locks it holds. The new manager should release all locks held by the old manager, since all operations undertaken by the old manager that are not yet complete must have been aborted.
- is duplicated by
-
SERVER-20018 After swapping catalog manager, clean up any distributed locks owned by the old catalog manager
- Closed
- related to
-
SERVER-21660 serverStatus can stall and trigger timeout error on SCCC distributed lock with v3.0 mongos and v3.2 cluster
- Closed