OpenShift Master: v3.9.0+ba7faec-1 Kubernetes Master: v1.9.1+a0ce1bc657 deployed via the branded Ansible playbooks
Metrics are installed as part of the ansible deployment via openshift_metrics_install_metrics=true and openshift_metrics_image_version=v3.9. The TTL value in hawkular-metrics pod is -Dhawkular.metrics.default-ttl=7, which, according to the documentation, should control the removal of any data from the Cassandra DB older than 7 days.
2+ months after the deploy I'm getting monitoring warnings that the Cassandra's PV usage grew over 20 Gb. I see that /cassandra_data directory inside the Cassandra pod contains fully loaded data folders going all the way back to the day of deployment - so, it appears that the Cassandra database is NOT automatically recycled as the documentation suggests it would.
Are there additional flags that need to be turned on? Really puzzling as I don't see anyone reporting similar problems in https://github.com/openshift/origin-metrics/issues