RE: memory related problems

From: Gil Cota <gil.cota_at_netcabo.pt>
Date: Wed, 17 Feb 2010 23:14:09 -0000
Message-ID: <054a01cab026$e9655ae0$bc3010a0$_at_cota@netcabo.pt>



Hi Helter,

That's because srvctl it routes commands through the root user , using sudo, as the root user does not have the HugePages configured,

that's why you are getting the swapping and stuff.

I got something similar but with the ulimits stuff , that my SO Admin limited the ulimit of root user, so every command with the srvctl didn't work.  

Just a test, try to configure the HugePages in the root user as well.  

--limits.conf

oracle hard memlock 25176064

oracle soft memlock 25176064

root hard memlock 25176064

root soft memlock 25176064      

I hope this will help,

Cheers

Gil  

From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org] On Behalf Of Helter Skelter
Sent: quarta-feira, 17 de Fevereiro de 2010 21:29 To: helter.skelterr_at_yahoo.com
Cc: oracle-l
Subject: Re: memory related problems  

sorry, I forget...

oracle 10.2.0.4

grid 11gR2

rhel5 x64  

cheers  


From: Helter Skelter <helter.skelterr_at_yahoo.com> To: oracle-l <oracle-l_at_freelists.org> Sent: Wed, February 17, 2010 10:26:30 PM Subject: memory related problems

Hi,  

I've got some strange memory related problems on my test environment.

Here is my configuration:  

System memory 32GB  

--oracle

sga_max_size 24G

sga_target 24G

pre_page_sga TRUE

pga_aggregate_target 4G  

--sysctl.conf

kernel.shmmax = 25316935680

kernel.shmall = 12361785

vm.nr_hugepages = 12293  

--limits.conf

oracle hard memlock 25176064

oracle soft memlock 25176064  

I've started oracle from sqlplus - everything works as expected:  

             total used free shared buffers cached

Mem: 32192 26620 5571 0 23 722

Swap: 20191 0 20191  

HugePages_Total: 12293

HugePages_Free: 4

HugePages_Rsvd: 0

Hugepagesize: 2048 kB  

Then, oracle autostart disabled ->reboot of server->sqlplus ->startup -> OK :)  

But...I've enabled autostart (srvctl), restarted server and then problems appears - a lot of swapping, high server load, oracle processes psp, mman, lmon at 70-80% cpu and in the end:

Process MMAN died, see its trace file

USER: terminating instance due to error 443  

after that I can with no problems start database with sqlplus, but when I stop it again and try start it with srvctl - the same...swappping, high load etc...  

any ideas where is the problem ?  

thanks            

--

http://www.freelists.org/webpage/oracle-l Received on Wed Feb 17 2010 - 17:14:09 CST

Original text of this message