This short report brings to light some interesting commits to scylla-cluster-tests.git master from the last week.
Commits in the d6fc1d7a…a56b93f8 range are covered.
There were 27 non-merge commits from 7 authors in that period. Some notable commits:
We updated cql-stress to the latest version containing a fix for creating proper tables based on workflow.
When we only have the owner ID and the AMI is not published in the marketplace (e.g., Oracle DB), we cannot use SSM to find it. We introduced a way to find them by owner and can use it by providing, e.g., resolve:owner:131827586825/x86_64/OL8.* for the ami_id_db_scylla parameter.
Scylla-bench version was bumped to v0.1.22 containing a fix for counter tables creation based on workflow.
A new job for specific test resources cleanup was added and can be easily used from Argus ‘Resources’ tab.
A new nemesis_grow_shrink_instance_type test parameter was introduced to control the instance type the grow-shrink nemesis uses. When this option is used, the shrink phase will decommission exactly the node that was added, leaving the cluster the same size as before nemesis.
Perf-simple-query results gained the perf_extra_jobs_to_compare Jenkins option so results can be compared across different jobs (releases).
We replaced pip and pip-compile with uv (a package installer written in Rust) which is much faster. This reduced the time to build out requirements.txt from 16s to less than a second (with a warm cache). See docs for instructions on how to use it.
See updated documentation (Notion) on how to configure AWS using OKTA in the local SCT git repo.
See you in the next issue of last week in scylla-cluster-tests.git master!