-
Type:
Task
-
Resolution: Unresolved
-
Priority:
Major - P3
-
Affects Version/s: None
-
Component/s: None
-
Replication
-
None
-
3
-
TBD
-
None
-
None
-
None
-
None
-
None
-
None
Currently we only briefly take/immediately release the global lock one time during setFCV (after the FCV is in the transitional FCV and before the FCV is in the final fully upgraded/downgraded state).
Operations can interleave with this lock in complex ways (see comment for more details). We should investigate if taking stronger locks (for example, taking this global lock barrier twice (once before each oplog entry write), or maybe taking the global lock during the whole operation) would help make this less complex and allow operations to serialize better with setFCV at least in terms of the locking.
- is related to
-
SERVER-105970 Operations that check feature flags with enable_on_transitional_fcv: true and write to disk can race with setFCV downgrade
-
- Open
-
-
SERVER-91269 Make feature flag checks less racy with setFCV
-
- Open
-
-
SERVER-102434 Investigate serializing setFCV so that FCV transitions happen in the same order with respect to other oplog entries on primary and secondaries
-
- Open
-
- related to
-
SERVER-106164 Add guardrails to ensure FCV upgrades/downgrades are finalized
-
- Backlog
-