This short report brings to light some interesting commits to scylla-cluster-tests.git master from the last week. Commits in the cb8e427b…d8d67bdb range are covered.
There were 38 non-merge commits from 13 authors in that period. Some notable commits:
The longevity-harry-2h test was added back to weekly runs, bringing it to regular tier1 triggers.
New performance scale-out/scale-in tests were added for elastic-cloud, checking performance under load (~50%) while scaling with different instance types. These manually triggered tests validate that there’s no performance regression during scaling operations.
Argus client was updated to version 0.15.5, improving logging and resource submission. Instance types are displayed in the Argus Resources tab. It’s helpful for tests with various DB instance types.
There was great effort this week to improve our unit tests through multiple commits. E.g. the test_coredump was fixed to use events fixture, setting up the events registry and eliminating warnings about unhandled exceptions.
Azure rate limiting was fixed with exponential backoff and error isolation, ensuring that temporary Azure API issues don’t prevent the entire cloud monitoring system from generating reports for other providers. This was a contribution by GitHub Copilot agent without much human intervention.
The cql-stress Docker tag was updated to v0.2.4 bringing rust driver 1.3.0 and support for extra_definitions in user profiles.
Rolling upgrade pipeline improvements were made: a missing provision stage was added and the pipeline was switched to use centralized runSctTest.
Scylla Cloud backend gained the ability to run stress loads, building upon the existing PoC by adding loader node instantiation capabilities. The longevity pipeline was also updated to support running SCT tests on Scylla Cloud backend in Jenkins.
See you in the next issue of last week in scylla-cluster-tests.git master!