Bug 702202

Summary: Higher system CPU utilization when running RHEL6 as guests than RHEL5.5
Product: Red Hat Enterprise Linux 6 Reporter: Kai <kai.xiang>
Component: kernelAssignee: Virtualization Maintenance <virt-maint>
Status: CLOSED WONTFIX QA Contact: Red Hat Kernel QE team <kernel-qe>
Severity: medium Docs Contact:
Priority: unspecified    
Version: 6.0   
Target Milestone: rc   
Target Release: ---   
Hardware: x86_64   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2013-04-30 21:49:58 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:

Description Kai 2011-05-05 03:42:50 UTC
Description of problem:

We are running kvm as vmm and upgrating our guest from RHEL5.5 to RHEL6.0
but we observer higher system utilization with same pressure/workloads.

One thing we notice is we have divider=10 in RHEL5.5 so the timer interrupt is less than what we got on RHEL6. 
to verify this, we use `cat /proc/interrupts |grep LOC; sleep 10; cat /proc/interrupts |grep LOC` to collect the interrupts inside 10s.
we found that the RHEL5 vm with divider=10. There is 100/s exactly for one cpu.
while with RHEL6 vm, it's ~883/s for one cpu.

we know the divider=10 will help for decrease system utilization. 
(exp: http://www.arrfab.net/blog/?p=62)

so I'm wondering if this is a expected? any similar tunnings for RHEL6 to lower the timmer interrupt as divider=10?


Version-Release number of selected component (if applicable):
6.0

How reproducible:
every time

Steps to Reproduce:
1. Setup two Guest VMM on KVM with RHEL6, one is RHEL5.5 one is RHEL6. add divider=10 for RHEL5.5
2. run some same workload for 2 VMM. (we use internal ones with constant pressure too complicate to setup and I think a simpler workload with ~50% load will be OK)
3. use `cat /proc/interrupts |grep LOC; sleep 10; cat /proc/interrupts |grep LOC` to collect the interrupts between two VMM.
  
Actual results:
Higher timmer interrupts/System utilization for RHEL6 then RHEL5.5+divider=10


Expected results:
Less or at leaset similar.

Additional info:

Comment 2 RHEL Program Management 2011-05-05 06:01:46 UTC
Since RHEL 6.1 External Beta has begun, and this bug remains
unresolved, it has been rejected as it is not proposed as
exception or blocker.

Red Hat invites you to ask your support representative to
propose this request, if appropriate and relevant, in the
next release of Red Hat Enterprise Linux.

Comment 3 Kai 2011-05-05 06:20:12 UTC
Thanks for the information.
One more question is if this also happended for 6.1 or newer version . will it be considerd? what is your suggest version number.
I can have a try with newer versions.

Comment 4 RHEL Program Management 2011-10-07 15:34:01 UTC
Since RHEL 6.2 External Beta has begun, and this bug remains
unresolved, it has been rejected as it is not proposed as
exception or blocker.

Red Hat invites you to ask your support representative to
propose this request, if appropriate and relevant, in the
next release of Red Hat Enterprise Linux.

Comment 5 Ademar Reis 2013-04-30 21:49:58 UTC
I'm closing this bug because it was reported long ago and we didn't hear any other reports of similar problems. If you can still reproduce it with recent RHEL6 releases (6.4.z), please reopen it. Thanks.