-
Type: Improvement
-
Resolution: Unresolved
-
Priority: Minor - P4
-
None
-
Affects Version/s: 6.0.16, 7.0.10, 8.0.0-rc4
-
Component/s: None
-
None
-
Query Execution
-
200
The dropIndex command generates an oplog entry on every shard containing the sharded collection. When clusterTime on 2 or more shards is identical, the corresponding change stream events will be indistiguishable including the resume token.
This does not make any problems when change stream events are consumed contiguously, however, if the stream is interrupted and resumed on the first of the identical events, the subsequent identical events will be skipped which might be unexpected by some event readers.
- is related to
-
SERVER-34554 Database drops do not have a total ordering in a change stream
- Backlog
-
SERVER-42008 Simultaneous sharded collection drops do not have a total ordering in change streams
- Backlog
- related to
-
SERVER-90266 Rename collection events do not have total ordering in change streams
- Backlog
-
SERVER-90091 Robustify change stream split event tests against identical events
- Closed