The Linux kernel automatically uses all available RAM for caching the file system to increase file access performance. Should the system require more RAM for processes, the memory for the buffer cache will automatically shrink as required. Looking at your output, I'd say there is nothing to worry about.
Btw, Oracle processes use shared memory, either Posix /dev/shm (AMM) or conventional (normal 4k or non-swap 2 MB hugepages). You can check it with the ipcs -m command and df /dev/shm. /dev/shm is 50 % of physical RAM by default, allocated only when used.
there are many times the CPU usage increases upto 100% usage and the swap space usage is empty..
There are two oracle database server (using for oracle application r12.1.3) sitting on this redhat server.Now only one db server is up and running since it CPU memory usage is going to 100 % we are unable to up the other one oracle server.
Please suggest how i can identify the root cause of this problem..
FYI - The problem is happening after we cloned our Oracle Application.
Fyi - when the CPU usage is high i used system moniter to check which process is using more CPU.
and did the following
system moniter->process (the process which is consuming more CPU memory) ->memory maps-> VM Size
Its showing 43917.1 GiB.
Is this a strange behavior or is this a normal one ?
And the process is pointing to the
Please advise on this ..Thanks
it should be your sga.
oracle processes maps sga. code, libraries it seems mapped files are also calculated.
VM size almost irrelevant to measure memory usage.
For example on the "top" command there is VIRT column.
VIRT -- Virtual Image (kb)
The total amount of virtual memory used by the task. It
includes all code, data and shared libraries plus pages that
have been swapped out.
What I meant to say is the process maps the sga,as the topic of this thread is about the ram usage of the process. Mapped Sga will vm size to be large, and ram usage will seem large, but no harm.
Edited by: Erman Arslan on 29.Ara.2012 01:52