This short report brings to light some interesting commits to scylla-cluster-tests.git master from the last two weeks.
Commits in the a8da7cf7…08f927dd range are covered.
There were 35 non-merge commits from 10 authors in that period. Some notable commits:
We reduced number of concurrent load threads in large partitions test to avoid IO/net limits and achieve better CPU utilization.
In K8s tests for getting logs we use
kubectl command, but it sometimes causes logs to be missing. We implemented logs retrieval with k8s python client with hope of better reliability.
Introduced 4 chaos-mesh failures: packet loss, corruption, network delay and limit bandwidth
to K8s tests.
Disabled speculative retry for performance tests because it could affect on max throughput.
Bumped the K8S version to 1.25.
Bumped default manager version to 3.1.
Since Ubuntu 18 version is EOL, we stopped testing it,
for tests that used it we switched to Ubuntu 22.
LogCollector now collects log of the manager’s scylla backend as sometimes is required for the investigations of manager bugs.
See you in the next issue of last week in scylla-cluster-tests.git master!