-
Type: Bug
-
Resolution: Fixed
-
Priority: Major - P3
-
Affects Version/s: 4.1.8
-
Component/s: Replication, Sharding, Testing Infrastructure
-
None
-
Fully Compatible
-
ALL
-
Repl 2019-03-11, Repl 2019-03-25
-
27
In continuous_stepdown.js, we assert that the replSetStepDown command always succeeds. However, it may fail if the node already stepped down after the call to getPrimary(), for example, because another node ran for election. Prior to SERVER-38516, we asserted that the command always threw (due to successful stepdown and connection closure) or failed (due to unsuccessful stepdown). After SERVER-38516, the command no longer throws due to successful stepdown, but it can still fail due to unsuccessful stepdown. We should not assert that the replSetStepDown succeeds.
It is possible that the replSetStepDown could throw if the node enters rollback after the call to getPrimary(), so we may still want to allow for this possibility.