Our DBAs recommend running repairs in Cassandra using the tool Reaper: Easy Repair Management for Apache Cassandra (cassandra-reaper.io)

Here is the proposal:

Because allCassandra nodes on all your 4 clusters have a JMX listener only listening on localhost, the setup of this tool could be:

1. cassandra-reaper.io/docs/usage/multi_dc_distributed/ with selection “Multiple Reaper instances with JMX access restricted to localhost” -> every node has one reaper process, which handles local repairs.
2. When you however allow JMX to listen on the ADMIN or PROD interface, then the setup can be easier “Operating Multiple DCs with a Single Reaper” -> centralized reaper process, which operates repair on multiple nodes. However this can be a security-concern as the JMX listener will listen on an external IP address.

What we would like to know from you:

- Is there any PEGA guideline to run repairs in Cassandra?

- Can we go ahead with above proposal?