Re: 12c RAC high CPU usage

From: Sanjay Mishra <>
Date: Mon, 30 Sep 2013 18:29:21 -0700 (PDT)
Message-ID: <>

Thanks Miller. I did now disabled it on both nodes. Suprisingly load is still high but only on Node 1. Rac alert log shows the following error every minutes [/u01/app/12.1/grid/bin/orarootagent.bin(4331)]CRS-5818:Aborted command 'check' for resource 'ora.drivers.acfs'. Details at (:CRSAGF00113:) {0:17:75} in /u01/app/12.1/grid/log/orarac1/agent/ohasd/orarootagent_root/orarootagent_root.log. 2013-09-29 21:16:45.661:
[/u01/app/12.1/grid/bin/orarootagent.bin(4331)]CRS-5014:Agent "/u01/app/12.1/grid/bin/orarootagent.bin" timed out starting process "/u01/app/12.1/grid/bin/acfsload" for action "check": details at "(:CLSN00009:)" in "/u01/app/12.1/grid/log/orarac1/agent/ohasd/orarootagent_root/orarootagent_root.log"

I did tried to stop and restart acfs using acfsload stop/sgtart but load is high on node 1

One thing is found that after stop and start of acfsload increase the CPU run queue while starting to 45+ but after it is started it is almost close to 10.In terms of high load, same process on top command like gipcd.bin, ohasd.bin.


 From: Seth Miller <> To: "" <> Sent: Monday, September 30, 2013 3:21 PM Subject: Re: 12c RAC high CPU usage  

Make sure you have disabled Transparent Huge Pages.


On Mon, Sep 30, 2013 at 1:22 PM, Sanjay Mishra <> wrote:

> Hi
> I had built the two node RAC using Virtual Box on Dell 2970 with 4G memory
> to each Node VM.
> when I tried to create Database, it basically hang at one point and stayed
> there for almost a day and put so much load in the server with topmost been
> gipcd.bin, lms, vtkm processes.
> I then finally cancelled and created the scripts and ran it and it created
> the database but after almost more than 18hr.
> Now as soon as I start the database, it shoots the CPU and not even able
> to logon to the server. I saw some site to use the following as mentioned
> by few sites but it just move the vtkm/lms process to low priotity but
> gipcd.bin is on high and system Run queue goes to 30+
> As soon as Database is stopped, the load/run queues reduce to almost 99%.
> selectksppstvl fromx$ksppi join x$ksppcv
> using(indx)whereksppinm='_high_priority_processes';KSPPSTVL
> --------------------------------------------------------------------------------LMS*|VKTM
> alter system set"_high_priority_processes"=''scope=spfile;
> Sanjay
> --


Received on Tue Oct 01 2013 - 03:29:21 CEST

Original text of this message