Script for monitoring Cassandra Cluster

Question

I need to report below metrics for a cassandra cluster at the end of the day

1. Cluster Load (Avg.)
2. No of read requests
3. No. of write requests
4. Read Latency
5. Write Latency
6. Long GC Pauses
7. No. of Connections. These have to be cluster wise and not node wise. Currently we are preparing the report through opscenter, which is a very manual process, I am planning to automate this task by writing a script. As I am new to Cassandra would like suggestions for where to begin with from experience folks here. Can this all be done using nodetool?

Thanks, MT

Answer 1

There are several possibilities here:

• If you use OpsCenter, you can use OpsCenter Metrics API to retrieve needed data. API allows to ask for data in given time range (start, end & end parameters), and then you can do any calculation on that data... The only catch is that you can't mix histogram data with gauges, etc.
• You can export DSE Metrics Collector's data via Prometheus, and then work with that data. There is a predefined config file for Prometheus, plus Grafana dashboards.