hi alalei! this error seem like caused by hbase config file(hbase-site.xml) is not loaded correctly by hadoop job "CubeHFileJob", then it use "localhost" for "hbase.zookeeper.quorum".
need more investigation the reason, this maybe a bug, howerver, if the task is assigned to nodemanager that co-deployed with zookeeper, then it can pass, I have try this, it works. (on my case, I have an test cluster, which is small, with four nodes, I deploy zookeeper on all nodemanagers, this is just trick, need solving it from root.) best regards! [email protected] From: alaleiwang Date: 2016-05-05 11:01 To: dev Subject: Re: CubeHFileJob fail with mr job configuration “2016-04-29 14:41:25,429 INFO [main] org.apache.zookeeper.ZooKeeper: Initiating client connection, connectString=localhost:2181 sessionTimeout=180000 watcher=hconnection-0x2888dbf40x0, quorum=localhost:2181, baseZNode=/hbase ” above was the hbase operation beginning;the reduce job was trying to get hbase meta table from zookeeper,while zookeeper host address should be defined in hbase-site.xml with parameter "hbase.zookeeper.quorum"; if nothing got from hbase.zookeeper.quorum(which could be loaded by Configuration conf = HBaseConfiguration.create(getConf())),localhost:2181 would be used,causing hbase operation failure(repeating trying to read hbase meta table from incorrect zookeeper host) i am not sure what i was thinking was all right... -- View this message in context: http://apache-kylin.74782.x6.nabble.com/CubeHFileJob-fail-with-mr-job-configuration-tp4331p4408.html Sent from the Apache Kylin mailing list archive at Nabble.com.
