-
Type:
Improvement
-
Resolution: Fixed
-
Priority:
Major - P3
-
Affects Version/s: None
-
Component/s: Live Restore
-
None
-
Storage Engines, Storage Engines - Foundations
-
StorEng - 2025-04-25
-
2
This ticket came out of the discussion started here.
There's a very small window during which live restore is in the background migration state but that state is only tracked in memory, and not yet persisted to the turtle file. If we crash during this period live restore will abort on a restart due to finding WiredTiger files in the destination, but the live restore state in the turtle file is NONE.
This ticket adds a comment explaining the above scenario and how users can recover from it.
- is related to
-
WT-14340 Make conn->flags atomic
-
- Closed
-
-
WT-14185 Remove workaround once copybara is enabled
-
- Closed
-
-
WT-14352 Create nested directory if directories do not exist during live restore file open and add test to simulate directory per db usage in MongoDB
-
- Closed
-
-
WT-13821 Review live_restore function comments
-
- Backlog
-
- related to
-
WT-14340 Make conn->flags atomic
-
- Closed
-
-
WT-14185 Remove workaround once copybara is enabled
-
- Closed
-
-
WT-14352 Create nested directory if directories do not exist during live restore file open and add test to simulate directory per db usage in MongoDB
-
- Closed
-