Last week in scylladb.git master (issue #259; 2024-12-08)

This short report brings to light some interesting commits to scylladb.git master from the last week. Commits in the 65949ce6078…f744007e136 range are covered.

There were 150 non-merge commits from 19 authors in that period. Some notable commits:


The topology coordinator will now detect tables that have shrunk, and merge adjacent tablets in order to meet the average tablet replica size goal.

An sstable scrub results in rewriting sstables, but did not remove the original sstable from metadata, causing a later reshape compaction to fail. This is now fixed.

Tablet repair operations are now tracked via the task manager.

The topology coordinator reads the request table, but if a new request arrives while the read is in progress, it could be missed. This is now fixed.

The decimal data type comparison operations have been improved not to cause stalls when comparing numbers with vastly different scales.

The data plane coordination code (“storage_proxy”) now uses host UUIDs to track hosts rather than network addresses. This simplifies the code and brings a nice performance improvement.

The scylla sstable dump-summary command now displays the tokens of the first and last keys. This helps associating an sstable with a node or a tablet.

Validation of sstable data checksums has been improved.

Alternator, ScyllaDB’s implementation of the DynamoDB API, now reports consumed WCU for item deletion operations.

The topology coordinator no longer waits for a replaced node to appear in gossip, since it may be gone already.

The gossip code will now clean up nodes that died before they could join Raft.


See you in the next issue of last week in scylladb.git master!