Hi everyone! Which function has to be used with each JMX Metric Type?
https://cassandra.apache.org/doc/latest/operating/metrics.html https://www.datadoghq.com/blog/how-to-collect-cassandra-metrics/ For example: to compute the read latency I did a ratio between ReadTotalLatency_Count JMX Counter and ReadLatency_Count JMX Timer, and the number corresponds to the one exposed via nodetool tablestats <keyspace>.<column_family> How should I consider the attributes: 95thPercentile, Mean etc... from ReadLatency bean org.apache.cassandra.metrics:type=Table,keyspace=test,scope=test_column_family,name=ReadLatency I found also a grafana open-source dashboard https://grafana.com/grafana/dashboards/5408 but I am not convinced about the displayed numbers if I compare with the numbers shown by the cassandra-stress tool. If I want the QPS does it make sense to use rate(WriteLatencyCount[5m]) in grafana ? The latency computed by the cassandra-stress-tool should almost match the latency shown by the JMX metrics or not? Which one do you monitor ClientRequest metrics or Table metrics or ColumnFamily? I am going to create my Grafana dashboard and explain how I configured it. Best, Sergio --------------------------------------------------------------------- To unsubscribe, e-mail: user-unsubscr...@cassandra.apache.org For additional commands, e-mail: user-h...@cassandra.apache.org