-
Type:
Bug
-
Resolution: Duplicate
-
Priority:
Major - P3
-
None
-
Affects Version/s: 8.3.0-rc0
-
Component/s: None
-
None
-
Catalog and Routing
-
ALL
-
CAR Team 2025-11-24
-
0
-
🟩 Routing and Topology
-
None
-
None
-
None
-
None
-
None
-
None
When a database is dropped via one router (mongos1), other routers (mongos0) retain stale routing information, causing subsequent collection operations to fail with NamespaceNotFound errors. This affects, for instance, createCollection and shardCollection commands but not insert operations, which handle retriability differently.
Steps to reproduce:
- Create a collection via router A
- Drop the database via router B
- Attempt to create/shard a collection via router A
- Operation fails with NamespaceNotFound
Expected: Router A should refresh its routing information, implicitly create the database and then retry the whole operation.
- is caused by
-
SERVER-103705 Explicitly use RouterRole on mongos commands
-
- Closed
-
- is duplicated by
-
SERVER-114129 Opt-in feature to implicitly create a database in the RouterRole API
-
- Closed
-