cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Matthew Allen <>
Subject Multi-site Active-Active replication - Maintenance Commands
Date Wed, 19 Mar 2014 03:42:05 GMT
Hi all,

We currently run a script that runs the following on our separate clusters
each night.  The script operates on 3 nodes (out of 9) a night to minimize
impact to the cluster.

  repair -pr

These separate clusters are going to start replicating between each other
within a few weeks.

I (may) have (mis)read that when using multiple data-centers, that a
"repair", instead of a "repair -pr" should be used.  I'm reluctant to do
this, as when we initially used to run a "repair", we would suffer hung
repairs and an increased load on the hosts due to multiple ranges being
repaired at the same time.

So going forward to a multi-dc cluster, we intend to run the following

  -local repair -pr

Is there a definitive answer for this ?



View raw message