-
Type:
Sub-task
-
Resolution: Fixed
-
Priority:
Major - P3
-
Affects Version/s: None
-
Component/s: Command Line Utilities
-
None
-
Storage Engines - Persistence
-
293.639
-
SE Persistence backlog
-
None
This is a sub-task to WT-17914, dependent on WT-17943.
This is the fix command to calibrate the database size, fix the cluster's issue described in WT-17563.
What we need for this task is to make sure the size fix is safe at runtime, and the basic logic here is derived from the verify function of __wt_verify_disagg_database_size.
- The database size should always be the sum of all collections' latest checkpoint sizes.
The timing for inject this logic is in __checkpoint_update_disagg_database_size which will be called at the end of each checkpoint. As the command call is in different thread to checkpoint, we can introduce a state machine for signal sync across them, and use the util command of fetch_database_size or grafana dashboard for living check the result.