As seen in the stack traces of BF-31930, there is a data race while reading the state of the _recipientCtx and updating the _recipientCtx.
Locking the _mutex (as done here) before reading the state of the _recipientCtx should fix the race.
- is caused by
-
SERVER-77639 Add new statistics for same-key resharding and index build timing during resharding
- Closed