Details
-
Bug
-
Resolution: Fixed
-
Major
-
6.5.0
-
Triaged
-
No
-
KV-Engine MH 2nd Beta
Description
There exists a race where we may have SyncWrite objects in the resolved SyncWrite queue and before we process them we call setReplicationTopology. This could free the pointers to the ReplicationChain objects that each SyncWrite holds and are used when attempting to complete them. This causes memcached to crash. This is exacerbated by the background completion of SyncWrites.
Attachments
Issue Links
- is duplicated by
-
MB-35552 TSan: data race on DurabilityMonitor::Position::lastWriteSeqno
- Closed