HBase scanner timeout during execution

[George Lu <lu...@gmail.com>]

Dear all,

I have three Drill nodes in cluster same as hdfs data nodes. The HBase
region servers are on that three nodes. I use Drill JDBC driver to submit
some count(*) group by queries. It failed and throw
"org.apache.hadoop.hbase.client.ScannerTimeoutException", the exception
came from one of the Drillbit.

I shut down two of the Drillbits and leave one in the cluster and execute
the Java program in that drillbit and finally it succeeded.

I know that error may related to HBase RPC communication and data transfer.
Well, my data transferred is not that much and the nodes are in the same
rack and supposed to me very fast in network communication.

No idea why that will happened?

Can anyone give me any idea considering Drill (potential data flow,
exception point), not only HBase.

Thanks!!!

George