Details
-
Improvement
-
Resolution: Works as Designed
-
Major - P3
-
None
-
None
-
None
-
None
-
Replication
Description
Currently, it's possible that if we write a jstests that kill a node and restart them before a stable checkpoint is taken. This leads to often unexpected initial syncs to happen when restarting the node because the WTRecoveryTimestamp will be null.