Hi,
I have 4 node hadoop/hbase setup where we run tasktracker on 3 of the
nodes.
I have given 2G to region servers and 1G for the child jvm. Whenever we run a
BuildTableIndex, during the map phase, there is cpu spike. Both the map task
and
the region server are high on CPU. I have dual core boxes with 8G RAM, and
usually the region server peeks at 50% and the mapper goes upto 150% cpu during
the job.
The reducer was also high on cpu which I brought down by adjusting
'hbase.index.max.buffered.docs'. What could be the reason for such a high cpu
usage during map phase ? Any fix or config tweak will be really helpful.
Thanks,
Murali Krishna
