-
Type: Bug
-
Resolution: Unresolved
-
Priority: Major - P3
-
None
-
Affects Version/s: 5.0.0, 6.0.0, 7.0.0, 8.0.0-rc0
-
Component/s: None
-
None
-
Catalog and Routing
-
CAR Team 2024-09-02, CAR Team 2024-09-30
-
3
SERVER-61416 changed the rename coordinator to be retried indefinitely when hitting any error after entering the blockCrudAndRename phase. The rationale behind the change was the following: since the rename is invoked in parallel on several shards, it would be problematic to fail the operation after it succeeded only on a subset of shards.
Even though that solution avoids introducing catalog inconsistencies on the from/to collections, we could find a more optimized way to safely abort the operation in case rename did not succeed on any shard and the state of from/to collections did not change under the critical section.