-
Type: Bug
-
Resolution: Unresolved
-
Priority: Major - P3
-
None
-
Affects Version/s: None
-
Component/s: None
-
Cluster Scalability
-
ALL
If the resharding recipient shard encounters a non retriable error in the cloning phase, it will inform the resharding coordinator so it will clean and abort the whole reshard operation.
The steps it will take are the following:
- RecipientService -> fails
- CoordinatorService -> starts aborting
- CoordiantorService -> sends abort to all participants
- RecipientService -> starts aborting
- RecipientService -> drops temp resharding collection
- RecipientService -> clears filtering metadata
- RecipientService -> starts async install of the filtering metadata
- RecipientService -> finishes installing filtering metadata with stale information by the async refresh
- CoordinatorService -> cleans up resharing temporary collection metadata
If (8) happens before (9), the recipient shard ends up with outdated filtering information for the temporary resharding collection.
- related to
-
SERVER-88978 Resharding coordinator should ensure participants have seen kAbort before dropping temp collection metadata
- Closed
-
SERVER-90224 Run sharding metadata consistency checks after ContinuousConfigShardTransition hook completes a transition
- Closed