cloudera-scm-server Process killed Automatically

Hi Dhanasekaran,

I don't see anything that indicates that the process is getting killed in
that log. Typically, when things get mysteriously killed, it's the Kernel
OOM killer. "dmesg" (or /var/log/kernel*) often has messages like "host
kernel: Out of Memory: Killed process 1234 (java)" in it. If you're seeing
those, then you're hitting memory pressure and you can add memory, run
fewer things, re-configure swap (has performance implications), adjust heap
sizes, etc.


-- Philip



On Sun, Sep 8, 2013 at 10:43 AM, Dhanasekaran Anbalagan
wrote:
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].

Hi Philip,

As per your suggestion. I don't find any OOM in dmesg and syslog.
By default scm-server process using 2G heap size. I don't want increase 2GB
to 3GB

/etc/default/cloudera-scm-server

can i add and test

#
# Java Options.
#
# Default value sets Java maximum heap size to 2GB, and Java maximum
permanent
# generation size to 256MB.
#

export CMF_JAVA_OPTS="-Xmx2G -XX:MaxPermSize=256m
-XX:+PrintTenuringDistribution -XX:+PrintGCDateStamps -XX:+PrintGCDetails
-XX:+PrintGC -Xloggc:/tmp/SCMGC/SCMgc.log -XX:+HeapDumpOnOutOfMemoryError
-XX:HeapDumpPath=/tmp/SCM"


Please correct me CMF_JAVA_OPTS variable.


-Dhanasekaran.





Did I learn something today? If not, I wasted it.

To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].