Are you missing data? Replication/recovery of blocks is automatic, and there isn't a manual process to it.
FWIW, for something like changing the hard drive configs on the box, it would be a better idea to unbalance the node ahead of time and then rebalance it. On Fri, Aug 9, 2013 at 7:18 AM, oc tsdb <[email protected]> wrote: > Hi, > > I am running a hadoop/hbase cluster with 6 nodes (1 name node, 5 data > nodes). > Two days back two of the machines went down due to some kernel crash. > These machines were rebooted after some time. > Meanwhile the cluster became unresponsive. > > Today, to create disk space and let other application services continue in > cluster, we had to wipe out hadoop data directory on one of the data nodes, > thinking that we should still be able to recover cluster as the replication > count was 3. > > Could you please point me to any procedure to recover data in my cluster. > > Thanks in advance. > > Regards, > OC. >
