Thanks Ted and Stack,
@Ted,  i also reproduced the OOM for scenario #1, and found the hints in the 
out file(it is 
/var/run/cloudera-scm-agent/process/*-hbase-REGIONSERVER/logs/stdout.log in 
CDH).
@Stack, it is not indeed hbase issue,  just because other application occupied 
more memory, then OS chose to killed RegionServer process.
Thanks for guys helps again.

Reply via email to