Hi Jean-Marc, we had some data in it. Just test data. I shut down the slave hbase cluster and deleted the cluster znode in zookeeper. Than I deleted the file in hdfs and started the hbase cluster. HBase generated a new hbase.id and the test data is still there.
Thank you very much. Replication is running again :-) Regard Hansi > Gesendet: Dienstag, 23. September 2014 um 14:09 Uhr > Von: "Jean-Marc Spaggiari" <[email protected]> > An: user <[email protected]> > Betreff: Re: how to change "hbase.id" > > Hi Hansi, > > This file is automatically creates by HBase. If there is no data on this > cluster, simply delete /hbase and ZK content and restart it, it will create > a new hbase.id. > > JM > > 2014-09-23 6:42 GMT-04:00 Hansi Klose <[email protected]>: > > > Hi, > > > > in our test environment I try to test the upgrade from cdh4.2.0 to cdh5.1.2 > > with replication. > > > > before the upgrade I copied the whole hbase cdh4 folder to from the master > > to the slave cluster > > and configured replication. This was working. > > > > After the upgrade I could not get the replication running again. > > > > The Problem seems to be that the hbase.id is the same on both cluster. > > > > The regionserver tells me: > > > > INFO org.apache.hadoop.hbase.replication.regionserver.ReplicationSource: > > Closing source 1 because: > > ClusterId xyz is replicating to itself: peerClusterId xyz > > > > I changed the hbase.id manual, but that does not work because the hbase > > master does not start. > > > > How can i change the hbase.id or create a new one? > > > > We run 2.3.0-cdh5.1.2 in our environment. > > > > Regards Hansi > > > > >
