Details
Description
When we recover the resharding ops in the drain mode we do a refresh of the collection. This refresh may hang here, because the in-memory critical section was already retaken and it will be released once the resharding operation is completed.
The proposed solution is calling this resharding-specific function instead of the generic one.
This issue was introduced when we replaced the in-memory critical sections by recoverable critical sections in SERVER-53653.
Attachments
Issue Links
- is related to
-
SERVER-53653 [Resharding] Take the critical section when renaming on recipient shards
-
- Closed
-
- related to
-
SERVER-56638 Fix flushReshardingStateChanges critical section race
-
- Closed
-