Details
-
Improvement
-
Status: Resolved
-
Low
-
Resolution: Fixed
-
None
-
None
Description
DeletedColumn and RangeTombstone use the local-timestamp to update the digest during repair.
Even though its only a second-precision timestamp, I think it still causes some differences in the merkle tree, therefore causing overrepair.
I attached a patch on trunk that adds a modified updateDigest() to DeletedColumn, which does not use the value field for its calculation.
Attachments
Attachments
Issue Links
- relates to
-
CASSANDRA-4905 Repair should exclude gcable tombstones from merkle-tree computation
- Resolved
-
CASSANDRA-4917 Optimize tombstone creation for ExpiringColumns
- Resolved
-
CASSANDRA-2698 Instrument repair to be able to assess it's efficiency (precision)
- Resolved