Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!

Bug 122942

Summary: xen-sources 2.6.12.6 - LVS kernel threads keep load above 1
Product: Gentoo Linux Reporter: Christopher G. Stach II <cgs>
Component: New packagesAssignee: Chris Bainbridge (RETIRED) <chrb>
Status: RESOLVED UPSTREAM    
Severity: normal    
Priority: High    
Version: 2005.1   
Hardware: All   
OS: Other   
URL: http://archive.linuxvirtualserver.org/html/lvs-users/2005-09/msg00031.html
Whiteboard:
Package list:
Runtime testing required: ---

Description Christopher G. Stach II 2006-02-15 10:05:41 UTC
I haven't yet located a patch, but there are potential ones in the thread.

PE2650 (Dual Xeon, HT enabled)

xen 3.0.0-r2
xen-sources 2.6.12.6
keepalived 1.1.11
ipvsadm 1.24
Comment 1 Chris Bainbridge (RETIRED) gentoo-dev 2006-02-15 11:14:11 UTC
Is this xen specific? It doesn't appear to be so. Has it been fixed in a later version of the vanilla sources? I don't really want to add random outside patches to the xensource release. It might be better to report it on the xen developers mailing list.
Comment 2 Christopher G. Stach II 2006-02-15 11:20:10 UTC
This is only specific to LVS.  It has nothing to do with Xen, but this kernel is one of the versions that the issue appears in.  I haven't tried recent vanilla or gentoo-sources because I don't run LVS on any of the non-xen machines.  Xen will be 2.6.16 eventually, so it may or may not not be worth investigating.  If the bug still exists in 2.6.16, then yes.  If it will be a long time before 2.6.16, then yes.
Comment 3 Chris Bainbridge (RETIRED) gentoo-dev 2006-04-10 06:05:40 UTC
xen-2.6.16 kernel is out now. You might like to try it and see if this bug still exists. The bug appears to affect all kernels, so it would be best to fix it upstream if there's still a problem.
Comment 4 Christopher G. Stach II 2006-04-11 16:09:03 UTC
Still have it in 2.6.16.
Comment 5 Chris Bainbridge (RETIRED) gentoo-dev 2006-04-12 02:26:36 UTC
Can you please try to get it fixed upstream? I would be hesitant to add a patch for this to the xen-sources, because it isn't xen specific, and it touches on the kernel internals which might have some unforeseen consequences...