refer to 0.20.6 version
this is a little different from standalone install ,as hbase will use pesudo/real cluster hadoop .so some diferences are listed below:
A.hbase-site.xml
1)hbase.rootdir
u should set it with the hadoop hdfs url,e.g. hdfs://url-to-hdfs:<port>/path-shared-by-regionserver
2)hbase.zookeeper.quorum
yes,we use a zk instance only managed by hbase,so set this to localhost is ok too.
B.starting sequence
1)start the hdfs and mapred daemons.note:as hbase will use MR to carry the row count etc,so mapred is a MUST also.
2)start-hbase
C.info after starting
1)use jps to check,it is differ to standalone:
8786 DataNode
9714 HMaster
9129 TaskTracker
8925 SecondaryNameNode
9656 HQuorumPeer
9000 JobTracker
8643 NameNode
the colorful pids are same with standalone,but additional ones are there also.
2) test zk by
echo ruok | nc localhost 2181
will return imok tip.
and the sequence of stoping is like this:
stop-hbase
stop-all //hdfs and maped
good luck!