-
Type:
Improvement
-
Resolution: Unresolved
-
Priority:
Major - P3
-
None
-
Affects Version/s: None
-
Component/s: Checkpoints
-
None
-
Storage Engines, Storage Engines - Persistence
-
SE Persistence backlog
-
None
We currently have crash points after checkpoint txn commits (before metadata updates and before metadata sync) as part of our deterministic crash testing infrastructure.
It would be worthwhile to also add a crash point before the transaction commit itself.