Last week in scylladb.git master (issue #206; 2023-11-26)

This short report brings to light some interesting commits to scylladb.git master from the last week. Commits in the eb674128ca…a472700309 range are covered.

There were 94 non-merge commits from 18 authors in that period. Some notable commits:

During commitlog replay, we skip over corrupted sections. However if the corrupted section also has corrupt size, it can lead to a crash. This is now fixed.

The bundled Prometheus node_exporter, used to expose operating system metrics, was updated to version 1.7.0 to address a security vulnerability.

In some cases a materialized view table’s schema can be constructed when the base table’s schema is not yet known. We now avoid this illegal state.

Previously, we added support for performing reconciliation (read repair) on partitions that are composed of a long sequence of tombstones without timeouts. This feature is now enabled only when the entire cluster is upgraded, to avoid warnings in the log.

For S3 backed keyspaces, the endpoint configuration is now checked when the keyspace is created, to avoid surprises later.

A bug in reconciliation (read repair) in conjunction with reverse queries and range tombstones, that could cause incorrect data to be returned from queries, has been fixed.

The sstable validation tools, scylla sstable validate-checksum and scylla sstable validate, now returns output in json format.

Node startup will now recalculate the schema digest fewer times during restart, resulting in faster startups.

Tablets are a new, experimental way of distributing data across nodes and shards. There is now support for updating drivers about tablet topology, which can change quite frequently.

When consistent cluster topology is enabled, we now reject a replace node operation if the node being replaced is not dead.

Repair has gained a new mode for small tables that makes repair significantly faster.

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