Details
-
Task
-
Resolution: Unresolved
-
Major - P3
-
None
-
None
-
None
-
None
-
Storage Execution
-
Execution Team 2024-02-19
Description
If a collection has replicated record ids, dbHash can scan in record id order rather than _id order. dbHash currently scans in _id order. For recordIdReplicated:true collections, we can go in natural order, as we already do for clustered collections.
Additionally, we should include each recordId when computing the hash, to make sure that the recordIds are indeed replicated.
In this ticket we need to handle just dbHash, not dbCheck.
This will also help with SERVER-86167 because then our tests can rely on dbHash for correctness.
Attachments
Issue Links
- depends on
-
SERVER-78348 Apply inserts on secondaries using record id
-
- In Code Review
-
- is duplicated by
-
SERVER-86212 Make dbHash be smart for recordIdReplicated:true collections
-
- Closed
-