-
Type:
Task
-
Resolution: Unresolved
-
Priority:
Major - P3
-
None
-
Affects Version/s: None
-
Component/s: None
-
None
-
Storage Execution
-
None
-
None
-
None
-
None
-
None
-
None
-
None
dbcheck_logs_minkey_maxkey.js writes to the secondary in standalone mode to introduce corruption, then asserts dbCheck's health log catches it with the expected minKey/maxKey entries. It was exempted from rid-enabled collections via overrideRecordIdsReplicatedFalse as the expedient fix for SERVER-118527. This ticket was the "decide and fix properly" follow-up.
Change
Delete the failpoint. Under the groupId design the exemption is no longer needed:
- dbCheck is rid-agnostic by design, so health log batch entries (minKey/maxKey) are unaffected by rids.
- The standalone write that introduces the corruption now also clears the secondary's groupId via C4 (SERVER-138309), which is the correct semantics for a node written outside replication. The test already sets TestData.skipCheckDBHashes, so the resulting out-of-group state does not trip fixture hash checks.
This resolves the original refactor question in the "make it compatible" direction: no rrid-specific handling is required because dbCheck never sees rids.
Acceptance criteria
- The test passes with the failpoint removed, in default and all-feature-flags variants.
- The corruption it introduces is still caught with the same minKey/maxKey health log entries.
- depends on
-
SERVER-138124 Membership-based recordId emission and honoring in the oplog path
-
- Open
-
- is depended on by
-
SERVER-138242 Retire the per-collection recordIdsReplicated option
-
- Open
-
- is related to
-
SERVER-138309 Clear the groupId on standalone document-data writes (C4)
-
- Open
-
- related to
-
SERVER-118527 Restore test coverage for dbcheck_logs_minkey_maxkey.js in the all feature flag tests variant
-
- Closed
-