Last week in scylla-cluster-tests.git master (issue #37; 2024-02-02)

This brief report highlights some fascinating commits to the scylla-cluster-tests.git master from the past week. The commits covered range from 818882db to 6833e66b.

During this period, 5 Software Developers in Test and 1 Software Developer made 22 non-merge commits. Here are some noteworthy changes:

We have started collecting and displaying syslog-ng stats in the monitor, enabling us to track logging rates and drops.

We’ve switched to ‘gp3’ EBS disks from ‘gp2’ for all K8S nodes we provision on EKS backend. These disks are both faster and more cost-effective.

We’re in the process of moving cloud cleanup procedures to the SCT codebase. The first step was to periodically remove unused Azure instances and Resource Groups. All cleanup scripts use the keep tag, where alive prevents deletion and an integer value specifies the number of hours to retain the instance or Resource Group.

Having recently started using ‘launch template’ resources in EKS, we now delete these too after test runs.

We’ve extracted some logic for creating a stress command, allowing for more widespread use.

We’ve added support for the cql-stress stress tool to SCT. This tool, similar to cassandra-stress but implemented in Rust, allows us to test rust-based drivers on a larger scale. Currently, we support read, write, counter_read, and counter_write commands. A new test based on this tool has been added. For more information, visit: GitHub - scylladb/cql-stress

Stay tuned for the next update on last week’s changes in scylla-cluster-tests.git master!