We are using kubernetes with opendistro. We started to notice an issue where /var would start filling up and tracked the issue to our pods that are running opendistro 1.9.0. The /tmp directory in the pod is between 16-30Gb in size and contains over 138 000 files with the name “metricsdb”:
Can you check the performance-analyzer.properties file located inside cat /usr/share/elasticsearch/performance-analyzer-rca/pa_config/performance-analyzer.properties (for the docker image) and see if the property, cleanup-metrics-db-files = true ?