ScyllaDB Manager 2.5

The ScyllaDB Manager team is pleased to announce the release of ScyllaDB Manager 2.5, a production-ready version of ScyllaDB Manager for ScyllaDB Enterprise customers and ScyllaDB Open Source users. ScyllaDB Manager is a centralized cluster administration and recurrent tasks automation tool.

ScyllaDB Manager 2.5 is focused on improving backup performance.

ScyllaDB Enterprise customers are encouraged to upgrade to ScyllaDB Manager 2.5 in coordination with the ScyllaDB support team.

The new release includes upgrades of both ScyllaDB Manager Server and Agent.

Useful Links:

Improvements

Backup start time improvement

Clusters having lots of backup files can suffer from a long wait time before uploading files. This is due to listing all remote files to calculate which files should be uploaded. The Upload would now start quickly with full speed as we changed to per file check.

Backup disk usage improvement

Uploaded backup files are immediately removed from a node storage, freeing disk space for compactions.

Agent memory usage optimisation

ScyllaDB Manager Agent with default settings consumes below 300MiB of RAM at peak, compare to more than twice in Manager 2.4

Agent pins to all interrupt CPUs

If ScyllaDB is configured to leave out interrupt CPUs the ScyllaDB Manager Agent pins to all of them instead of just the last. This update allows a more fair core distribution and decreased CPU load. Note Agent resources are capped using systemd slice.

Sctool status command shows errors information for failed checks

Example of failed CQL check

Other changes

  1. Support for v1 backups, created by  ScyllaDB Manager version 2.0 and not migrated, is removed.
  2. Backup listing implementation processes backup manifests sequentially, this improves server memory usage and prevents OOM crashes.
  3. TLS/HTTPS self-signed certificates are automatically generated on server start if not set. The scyllamgr_ssl_cert_gen script is removed.
  4. Startup errors (like wrong configuration) are visible in the service status and journal logs. Now it’s only visible in journalctl -xe.
  5. Swagger UI and migration files are now embedded in the server binary /etc/scylla-manager containing only the config file.
  6. Fixes a bug in Manager health check functionality, which causes a deadlock, which in turn causes Prometheus endpoint hangs and ends up causing File Descriptor starvation (part of ScyllaDB Manager 2.5.1)

30 Aug 2021