Last week in scylladb.git master (issue #220; 2024-03-03)

This short report brings to light some interesting commits to scylladb.git master from the last week. Commits in the 57c408ab5d…94cd235888 range are covered.

There were 47 non-merge commits from 9 authors in that period. Some notable commits:

Yet another case of repair failures if a table is dropped during repair was fixed.

A potential data resurrection problem when cleanup was performed as a side-effect of regular compaction was fixed.

A use-after-free in topology coordinator error handling was fixed.

A data race in the service that propagates cache hit rate statistics across the cluster was fixed.

We now lock raft group 0 while taking a snapshot (to populate another node) to preserve atomicity.

The totimestamp() CQL function is now protected against undefined behavior when applied to extreme values.

As a side effect of work to reduce test run-time, there is now a config item that can be used to change the internal page size.

We now no longer emit a removed node notification to drivers when a node IP address changes. The notification confused drivers, causing then to lose the ability to reconnect.

An internal RPC for schema changes, which may only run on shard 0, is now enforced to do so.

A deadlock when adding or removing nodes to Raft was fixed.

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