after upgrading the stack from 5.2.1 to cdh 5.3.0 hbase stopped working. for example when in the hbase shell trying to scan a table it just hangs until it timeouts. here is is a jstack of the hbase shell client: http://pastebin.com/dwfi7CwP
my hbase config is simply:
<value>node00, node01, node02</value>
running two hbase masters...
any idea what's causing it and if there is a workaround?
fyi if I just downgrade hbase to cdh 5.2.1 while leaving the rest of the cluster (all hadoop, zookeeper etc) on cdh 5.3 it works as charm so it is something to do with hbase.
Are both HDFS and ZooKeeper running?
Are you able to do hdfs dfs command over HDFS?
Also please see if znodes for HBase exist in zookeeper shell; zookeeper-client --> ls /hbase
If you created the cluster using Cloudera Manager, make sure you did deploy client configuration.
yes, everything is running - whole cluster on cdh 5.3 and hbase is stuck. when downgrading _just_ hbase to cdh 5.2.1 (everything else stays inc config) it *works*. the problem occurs for all cdh 5.3 hbase releases.