This short report brings to light some interesting commits to scylla-cluster-tests.git master from the last week.
Commits in the 528b0a52…df2b91cf range are covered.
There were 38 non-merge commits from 15 authors in that period. Some notable commits:
Nemeses target only data nodes by default. Since most also apply to zero-token nodes (including config and schema changes), both data and zero-token nodes are now explicitly set as nemesis targets.
A new job was added with four zero-token nodes to run all supported nemeses.
All pip dependencies were updated.
A templating mechanism for monitoring dashboards was introduced, keeping SCT in sync with changes in the Scylla Monitoring repo.
Gemini now supports log compression using ZSTD, allowing longer runs without running out of disk space.
Sanity jobs and 90% utilization tests now use simulated racks for a more realistic testing environment.
Due to a Grafana renderer issue, monitoring was reverted to version 4.8.
Performance test step durations can now be configured using the perf_gradual_step_duration
parameter, making it easier to modify predefined step lengths. This was used in a new Java driver performance test.
The modify_table_twcs_window_size
nemesis was refactored and now tests TWCS borders across wider ranges.
Performance regression tests now respect the pre_create_keyspace
parameter, allowing easier configuration of the initial number of tablets.
See you in the next issue of last week in scylla-cluster-tests.git master!