-
Type: Bug
-
Resolution: Unresolved
-
Priority: Major - P3
-
None
-
Affects Version/s: None
-
Component/s: None
-
None
-
Replication
-
ALL
This is a very corner case. But since we ignore duplicateKey during initial sync and during oplog application, I think the following sequence is possible:
Sync Source | Syncing nodes |
---|---|
Insert 1, 2: [1, 2] |
Clone 1, 2: [1, 2] |
Rollover/delete 1: [2] |
|
Insert 1, 3: [2, 1, 3] |
Clone 1, 3 [1, 2, 3] because duplicateKey on (1) is ignored |
Start oplog application | |
Apply delete(1), insert(1), insert(3): [2, 3, 1] because duplicateKey on (3) is ignored |
- is related to
-
SERVER-66775 Enhance consistency checking for capped collections in testing
- Backlog