This short report brings to light some interesting commits to scylla-cluster-tests.git master from the last week. Commits in the 3418fb9c…50020224 range are covered.
There were 24 non-merge commits from 10 authors in that period. Some notable commits:
There’s a new Scylla Manager installation test on the Rocky9 distribution.
Download, load&stream bandwidth metrics for Scylla Manager restore tests are now sent to Argus, enhancing the informativeness of restore benchmark graphs.
The default Scylla Manager version was updated to 3.4.
Instance provisioning fallback to on demand is now enabled by default in the AWS backend, ensuring SCT retries with on-demand instances when spot instance creation fails.
The Latte tool version was bumped to 0.28.1-scylladb, allowing “batch” query testing.
A new predefined throughput steps with tablets test pipeline was added and scheduled for weekly runs. This test verifies latencies across different throughput steps and the maximum throughput Scylla can achieve. Additionally, write workload test was improved by running c-s directly in the loader OS instead of Docker, enhancing stability.
The mgmt_cli_test.py module was refactored to improve organization, naming, and test reusability for Scylla Manager.
Dependabot will now track Docker-based loaders. This required updates to configuration files but will not impact their usage in tests.
See you in the next issue of last week in scylla-cluster-tests.git master!