-
Type:
Bug
-
Resolution: Unresolved
-
Priority:
Major - P3
-
None
-
Affects Version/s: None
-
Component/s: Sharding
-
None
-
Storage Execution
Currently, for testing purposes, the recordIdsReplicated option is being removed automatically on downgrade. This was done as part of SPM-2983. The current implementation does not take into account any sharding guarantees, like stopping migrations, holding DDL locks, or honoring critical sections. This leaves the possibility to have inconsistent metadata across shards temporarily.
WRITING-25576 describes a correct procedure to update catalog metadata on sharded clusters. The purpose of this ticket is to follow that design to prevent having metadata inconsistencies when removing recordIdsReplicated on downgrade in a sharded cluster.
- is caused by
-
SERVER-89633 Create a new builder for recordIdsReplicated:true
-
- Closed
-
- is duplicated by
-
SERVER-87931 MovePrimary + FCV upgrade / downgrade race may dodge FCV cleanup & checks
-
- Closed
-
- related to
-
SERVER-87931 MovePrimary + FCV upgrade / downgrade race may dodge FCV cleanup & checks
-
- Closed
-
-
SERVER-88238 checkMetadataConsistency interleaves with collMod during upgrade / downgrade
-
- Closed
-
-
SERVER-89634 Enable tests that concurrently perform DDL and setFCV operations on the recordIdsReplicated:true variant
-
- Closed
-