-
Type:
Task
-
Resolution: Won't Fix
-
Priority:
Major - P3
-
None
-
Affects Version/s: None
-
Component/s: Checkpoints
-
None
-
Storage Engines, Storage Engines - Persistence
-
SE Persistence - 2026-02-13
-
None
Following on from WT-16641's findings, we should lift up the b-tree size computation into the b-tree layer if possible. This will mean that the block manager no longer owns size tracking which will be a nice improvement.
Scope:
- Investigate if this is indeed possible, if we do not have the relevant block metadata at the b-tree layer then we will need to continue with the current implementation. Regardless the second piece of this ticket.
- Define 2 functions, b-tree size increase, b-tree size decrease. Hook into the relevant paths, either up in b-tree or down in block manager.
- Test
- is related to
-
WT-16641 bytes_total leak in disagg during 1-for-1 page swap
-
- Closed
-