|
Hi, HBase users. I'm working with a 2 Nodes HDFS+HBASE cluster in full distributed mode for 2 months with success now (TABLE with more than 1.000.000.000 rows) . Everything works fine, I used also PHOENIX code from Saleforces with success. BUT. When I added a new node. HBase does not work for the third one. 2 Nodes : HDFS 1.2.1 HBase 0.94.12 OS: Ubuntu Server 12.04.3 The new node is on Centos 6.4 (same version and conf files but different OS) The HDFS part works fine (this node is already part of the HDFS cluster) Ganglia monitoring works well with this centos node The pb is very odd. When I start region server and zoo keeper on third node, I can see it : [hduser@nc-h04 ~]$ jps 13275 HRegionServer 13445 Jps 13383 HQuorumPeer 1499 DataNode I checked client ports (zoo keeper port and 60010) and access is ok and daemon is running. BUT there are no Web UI for third region server. And one last odd thing: in Web UI of the master I see strange report. The master see the region server but with wrong values for heap : nc-h04,60020,1383565134348 Mon Nov 04 12:38:54 CET 2013 requestsPerSecond=0, numberOfOnlineRegions=0, usedHeapMB=0, maxHeapMB=0 I checked logs but the only error I see is that : 2013-11-04 13:03:54,510 DEBUG org.apache.hadoop.hbase.io.hfile.LruBlockCache: Stats: total=1.98 MB, free=239.7 MB, max=241.69 MB, blocks=0, accesses=0, hits=0, hitRatio=0, cachingAccesses=0, cachingHits=0, cachingHitsRatio=0, evictions=0, evicted=0, evictedPerRun=NaN 2013-11-04 13:04:10,789 INFO org.apache.hadoop.hbase.HBaseFileSystem: Create Writer, sleeping 1000 times 6 2013-11-04 13:04:16,790 INFO org.apache.hadoop.hbase.util.FSUtils: FileSystem doesn't support getDefaultReplication 2013-11-04 13:04:16,790 INFO org.apache.hadoop.hbase.util.FSUtils: FileSystem doesn't support getDefaultBlockSize Anybody had same problem with Centos ? Any advice ? Why the daemon is running but there are no Web UI ???? I'm running out of solution. Must back to Ubuntu Server? CDH works with Centos, doesn't it? Regards, --
Damien CAROL
|
