-
Type: Task
-
Resolution: Fixed
-
Priority: Major - P3
-
Affects Version/s: None
-
Component/s: Sharding
-
Fully Compatible
-
Sharding 2018-11-19, Sharding 2018-12-03
If coordinateCommit gets retried due to a network error but the first coordinateCommit call succeeded in committing, we need to make sure that we recover the correct decision so that it may be returned to mongos. This overlaps with the work to be able to recover the commit decision in case of mongos failover, and potentially could replace/remove the need for it.
- depends on
-
SERVER-38095 Remove unused OperationSessionInfoFromClient argument to OperationContextSessionMongod constructor
- Closed
- is depended on by
-
SERVER-37344 Implement recovery token for retrying a commit command on a different mongos
- Closed