Ted, I guessed the problem could be due to only single zookeeper server in hbase.zookeepr.quorumpeer. So, i have added the region server machine also apart from the master. Now, i don't see any such FAIL cases as mentioned below. (which was the case earlier)
Handling transition=RS_ZK_REGION_ FAILED_OPEN, server=vamshi_RS,60020,1377265388499, region=2dc3d895b55455dd06880fb0f15bb80d But, still i don't see the regions moving to other server i.e region server machine. I have created 2 tables on master machine and stored few millions of rows in each. Through hbase master web UI (http://<host>:60010), i see Region Servers ServerNameStart timeLoad vamshi,60020,1377502384119 <http://vamshi:60030/>Mon Aug 26 13:03:04 IST 2013requestsPerSecond=0, numberOfOnlineRegions=4, usedHeapMB=170, maxHeapMB=991 vamshi_RS,60020,1377502387395<http://vamshi_RS:60030/>Mon Aug 26 13:03:07 IST 2013requestsPerSecond=0, numberOfOnlineRegions=0, usedHeapMB=24, maxHeapMB=991 Total: servers: 2 requestsPerSecond=0, numberOfOnlineRegions=4 Why is it so..? I have set the dfs.replication to 3. Is this below statement not true..? Irrespective of the cluster size, the number of regions (data) should be spread uniformly among all the nodes in the cluster . Which implies, here in above case, "4 regions on server 'vamshi' and zero regions on server vamshi_RS " SHOULD BE CHANGED automatically to "two regions on server 'vamshi' and two regions on server vamshi_RS" .. Is my understanding right here? hbase-site.xml: <property> <name>hbase.rootdir</name> <!--value>hdfs://vamshi:54310/home/biginfolabs/BILSftwrs/hbase-0.94.10/data/</value--> <value>/home/biginfolabs/BILSftwrs/hbase-0.94.10/hbstmp/</value> </property> <property> <name>hbase.cluster.distributed</name> <value>true</value> </property> <property> <name>hbase.master</name> <value>vamshi</value> </property> <property> <name>hbase.zookeeper.quorum</name> <value>vamshi,vamshi_RS</value> </property> <property> <name>hbase.zookeeper.property.dataDir</name> <value>/home/biginfolabs/BILSftwrs/hbase-0.94.10/zkptmp</value> </property> On Mon, Aug 26, 2013 at 2:12 PM, Andrew Purtell <[email protected]> wrote: > Two nodes is insufficient. Default DFS replication is 3. That would be the > bare minimum just for kicking the tires IMO but is still a degenerate case. > In my opinion 5 is the lowest you should go. You shouldn't draw conclusions > from inadequate deploys. > > On Friday, August 23, 2013, Vamshi Krishna wrote: > > > Hello all, > > I set up a 2 node hbase cluster. I inserted million rows in > to > > hbase table. Machine1-master and region server > > machine-2-only region server. > > After the insertion completed, Still i see the 4 regions formed for the > > inserted table are residing on the machine -1 only. I waited more than 5 > > minutes, as the default balancer period is 5min from hbase apache > > documentation. But, still i notice all the 4 regions of the table are on > > machine-1 only. not distributed automatically. > > I even enabled the balancer from shell using "balance_switch true" > but > > no result even after 5 min. > > > > Could any one tell me why is it happening so.? > > > > > > -- > > *Regards* > > * > > Vamshi Krishna > > * > > > > > -- > Best regards, > > - Andy > > Problems worthy of attack prove their worth by hitting back. - Piet Hein > (via Tom White) > -- *Regards* * Vamshi Krishna *
