Details
-
Bug
-
Resolution: Fixed
-
Major - P3
-
None
-
None
-
Fully Compatible
-
ALL
-
Repl 2019-01-14
-
45
Description
When enableMajorityReadConcern='false', the stable optime can be ahead of the last committed optime, so the replication coordinator updates the committed snapshot based on the last committed optime instead of the stable optime. This causes a crash if the last committed optime is null, since _updateCommittedSnapshot_inlock() expects a non-null optime. We should not call _updateCommittedSnapshot_inlock() when the last committed optime is null.