Details
-
Bug
-
Status: Resolved
-
Major
-
Resolution: Fixed
-
None
-
None
Description
We currently base our decision to retry the index rebuild on the INDEX_DISABLE_TIMESTAMP changing. This works fine when we disable the index on a write failure, because the transition from DISABLE -> INACTIVE will fail and we'll try the rebuild again later. However, this is not the case when the index is left ACTIVE. As an additional safeguard, we should fail the rebuild and retry again in the next polling.