Re: memory related problems
- From: Helter Skelter <helter.skelterr@xxxxxxxxx>
- To: Yong Huang <yong321@xxxxxxxxx>
- Date: Thu, 18 Feb 2010 23:54:44 -0800 (PST)
Hi.
When I start srvctl HugePages weren't used and there were no other instances.
Anyway, problem solved.
It was bug:
Due to bug 9084067 (fixed in 11.2.0.2), ohasd.bin and oraagent.bin can not use
ulimit settings properly upon automatic startup of Grid Infrastructure when
node restarts:
ID 983715.1
thanks, cheers
________________________________
From: Yong Huang <yong321@xxxxxxxxx>
To: helter.skelterr@xxxxxxxxx
Cc: oracle-l@xxxxxxxxxxxxx
Sent: Thu, February 18, 2010 6:53:44 PM
Subject: Re: memory related problems
Your SGA, memlock and hugepage size are too close to each other. For
reason I don't know, sometimes hugepages have to be a little larger than
SGA. In any case, memlock can be set to very large with no harm. It
doesn't actually allocate memory as hugepages does.
When you start using srvctl and have a lot of swapping, is HugePages
used? Check /proc/meminfo. You probably get out of memory error in
$ORACLE_HOME/log/<host>/racg/imon_<db>.log.
You don't another instance on this box including ASM, do you?
Yong Huang
Other related posts:
