Thanks Jonathan, Aaron, Daniel! I have a related question. I would like to get a copy of data from these 12-server cluster with manually assigned babanced server tokens, and set it up on a new cluster. I would like to minimize the number of the server on the new cluster without having to build 12 servers on new cluster, copy snapshots from old cluster to new cluster, and start de-commission some servers on new cluster to get it down to the desired number of servers on the new cluster. I am OK with missing data since the last repair ran on the old cluster. Would the following data copy strategy work?
Create a new cluster with 4 servers and manually assign balanced server tokens to these four servers; copy most recent snapshot from every 3rd server from the old cluster (of 12 servers) and put them on the new cluster in the same order as they were from old cluster. Run repair on every other node or all nodes on new cluster. Thanks! Huy On Tue, Mar 15, 2011 at 5:16 PM, aaron morton <aa...@thelastpickle.com>wrote: > AFAIK you should run it on every node. > > http://wiki.apache.org/cassandra/Operations#Repairing_missing_or_inconsistent_data > > > <http://wiki.apache.org/cassandra/Operations#Repairing_missing_or_inconsistent_data> > Aaron > > On 16 Mar 2011, at 06:58, Daniel Doubleday wrote: > > At least if you are using RackUnawareStrategy > > Cheers, > Daniel > > On Mar 15, 2011, at 6:44 PM, Huy Le wrote: > > Hi, > > We have a cluster with 12 servers and use RF=3. When running nodetool > repair, do we have to run it on all nodes on the cluster or can we run on > every 3rd node? Thanks! > > Huy > > -- > Huy Le > Spring Partners, Inc. > http://springpadit.com > > > > -- Huy Le Spring Partners, Inc. http://springpadit.com
