-
Type:
Bug
-
Resolution: Fixed
-
Priority:
Major - P3
-
Affects Version/s: 9.0.0-rc0, 8.3.0
-
Component/s: Sharding, Upgrade/Downgrade
-
None
-
Catalog and Routing
-
Fully Compatible
-
CAR Team 2026-06-08, CAR Team 2026-06-22
-
🟩 Routing and Topology
-
None
-
None
-
None
-
None
-
None
-
None
Currently, the condition for running shard server phase1 tasks is role->has(shard). We should consider changing this to role->isShardOnly instead.
This is because config servers are also shards so it can end up executing the phase1 logic twice (2nd time here). And some logic that was intended to run on config first like the changes in SERVER-102971 to move abort resharding logic end up being useless because the shard1 phase will end up being executed first.
- is caused by
-
SERVER-119476 Resume setFCV from the point where it got interrupted
-
- Closed
-
- is related to
-
SERVER-104789 Resharding fails to abort on setFCV on a cluster with config shard
-
- Backlog
-
-
SERVER-102971 Abort resharding before draining DDL coordinators during FCV downgrades
-
- Closed
-
-
SERVER-127574 Write setFCV upgrade/downgrade test with pinned resharding version
-
- Closed
-