-
Type: Bug
-
Resolution: Fixed
-
Priority: Major - P3
-
Affects Version/s: None
-
Component/s: Sharding
-
Fully Compatible
-
ALL
-
Sharding 2021-04-05
-
165
The current implementation of the onCompletion callback on the instances of the ShardingDDLCoordinator are simply returning if there was a NotPrimaryError or ShutdownError category, if there is a write and the DDL operation fails because of a stepdown, then that error would not propagate to the router nor the operation retried, giving an ok:0 result to the user when actually the operation haven't finished. We should throw on those categories but leaving the current approach of not removing the coordinator document on those scenarios.
- is caused by
-
SERVER-54587 Make create collection resilient to stepdowns
- Closed
-
SERVER-54605 Make rename collection coordinator resilient to stepdowns
- Closed
-
SERVER-54944 Make drop collection resilient to stepdowns
- Closed
-
SERVER-54945 Make drop database resilient to stepdowns on sharded cluster
- Closed