Bug 679905

Summary: CPU scheduler load balancer is broken (cgroups)
Product: Red Hat Enterprise Linux 6 Reporter: Rik van Riel <riel>
Component: kernelAssignee: Rik van Riel <riel>
Status: CLOSED UPSTREAM QA Contact: Red Hat Kernel QE team <kernel-qe>
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: 6.2CC: bburns, lwang, lwoodman, tburke, vgoyal
Target Milestone: rc   
Target Release: ---   
Hardware: All   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: 672113 Environment:
Last Closed: 2011-12-12 17:22:14 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Bug Depends On: 672113    
Bug Blocks: 767187    

Description Rik van Riel 2011-02-23 20:11:41 UTC 
+++ This bug was initially created as a clone of Bug #672113 +++

Description of problem:

The CPU scheduler's load balancer in RHEL 6 is broken. With my PLE test, it was common for 1 4-CPU virtual machine to get around 300% CPU time, while the other 4-CPU virtual machine (contending for the same 4 CPUs) only got about 100% CPU time.

Upstream has several patches to fix load balancing.  I would like to try backporting those to RHEL 6 to see if that resolves the issues observed.

Note that CPU load balancing with cgroups will probably still be broken, even with those changesets applied; that is another bug alltogether.

(This bug is to track not-yet-materialized cgroups load balancing fixes from upstream)