-
Type:
Task
-
Resolution: Fixed
-
Priority:
Major - P3
-
Affects Version/s: None
-
Component/s: Block Manager
-
Storage Engines, Storage Engines - Persistence
-
SE Persistence backlog
-
None
We currently treat page discards as a full page write of a tombstone, but we should instead treat it more like a delta write.
The Server team has requested that we set the LSNs for a tombstone write accordingly:
- base_lsn = the LSN of the last full page
- backlink_lsn = the LSN of the last page version
Note that this would require us to store base_lsn in the address cookie, so that this information would be available for fast truncate as explained in WT-14960.
- is depended on by
-
WT-14960 Review what the address cookie should contain
-
- Closed
-
- is related to
-
WT-15158 __layered_update_gc_ingest_tables_prune_timestamps WT_ASSERT(session, prune_timestamp >= btree->prune_timestamp) failing while running hello_with_standby.js
-
- Closed
-
-
WT-15003 Remove the WT_DELTA_CELL_INT structure and pack internal page deltas with WT_CELL
-
- Open
-
-
WT-15191 Write a regression test for WT-15158
-
- Closed
-
- related to
-
WT-14365 Create performance tests to measure eviction efficiency
-
- Closed
-
-
WT-14988 Implement get_page_id function in PALM
-
- Closed
-