-
Type: Bug
-
Resolution: Fixed
-
Priority: Major - P3
-
Affects Version/s: None
-
Component/s: Replication
-
None
-
Fully Compatible
-
ALL
-
Repl 2018-06-18
If a commitTransaction command is successful but times out waiting for replication, or the response to the driver is lost, the driver may retry the commitTransaction command. In that case, the retried commitTransaction command will not do any writes, thus it will not advance the lastOpTime on the Client, nor will it take the Global lock at all to trigger the GlobalLockAcquisitionTracker. On commitTransaction retries we should call ReplClientInfo::setLastOpToSystemLastOpTime, to ensure that we wait for write concern appropriately.
- related to
-
SERVER-27067 Some Commands do not wait for write concern for no-op writes
- Closed
-
SERVER-33475 Retried findAndModify doesn't properly wait for writeConcern
- Closed