We are noticing nodetool repair processes are not completing after a weeks worth of time, and have resulted in some Cassandra nodes having more than one process running do to cron scheduled. We are also chasing some performance degradation after upgrading all nodes to version 1.2.8 last Friday and would like to resolve this multiple repairs running at once issue in an effort to troubleshoot our performance issues.

We'd like to know more about what is happening with the repair option. Is there a way to gracefully terminate them or any adverse affect to killing the processes we should look out for?

Andy L