[Users] 6.4 CR: oVirt 3.1 breaks with missing cpu features after update to CentOS 6.4 (6.3 + CR)

[Dan Kenigsberg]

On Tue, Mar 05, 2013 at 12:32:31PM +0100, Patrick Hurrelmann wrote:
> On 05.03.2013 11:14, Dan Kenigsberg wrote:
> <snip>
> >>>>
> >>>> My version of vdsm as stated by Dreyou:
> >>>> v 4.10.0-0.46 (.15), builded from
> >>>> b59c8430b2a511bcea3bc1a954eee4ca1c0f4861 (branch ovirt-3.1)
> >>>>
> >>>> I can't see that Ia241b09c96fa16441ba9421f61a2f9a417f0d978 was merged to
> >>>> 3.1 Branch?
> >>>>
> >>>> I applied that patch locally and restarted vdsmd but this does not
> >>>> change anything. Supported cpu is still as low as Conroe instead of
> >>>> Nehalem. Or is there more to do than patching libvirtvm.py?
> >>>
> >>> What is libvirt's opinion about your cpu compatibility?
> >>>
> >>>      virsh -r cpu-compare <(echo '<cpu match="minimum"><model>Nehalem</model><vendor>Intel</vendor></cpu>')
> >>>
> >>> If you do not get "Host CPU is a superset of CPU described in bla", then
> >>> the problem is within libvirt.
> >>>
> >>> Dan.
> >>
> >> Hi Dan,
> >>
> >> virsh -r cpu-compare <(echo '<cpu
> >> match="minimum"><model>Nehalem</model><vendor>Intel</vendor></cpu>')
> >> Host CPU is a superset of CPU described in /dev/fd/63
> >>
> >> So libvirt obviously is fine. Something different would have surprised
> >> my as virsh capabilities seemed correct anyway.
> > 
> > So maybe, just maybe, libvirt has changed their cpu_map, a map that
> > ovirt-3.1 had a bug reading.
> > 
> > Would you care to apply http://gerrit.ovirt.org/5035 to see if this is
> > it?
> > 
> > Dan.
> 
> Hi Dan,
> 
> success! Applying that patch made the cpu recognition work again. The
> cpu type in admin portal shows again as Nehalem. Output from getVdsCaps:
> 
>    cpuCores = 4
>    cpuFlags = fpu,vme,de,pse,tsc,msr,pae,mce,cx8,apic,sep,mtrr,pge,
>               mca,cmov,pat,pse36,clflush,dts,acpi,mmx,fxsr,sse,sse2,
>               ss,ht,tm,pbe,syscall,nx,rdtscp,lm,constant_tsc,
>               arch_perfmon,pebs,bts,rep_good,xtopology,nonstop_tsc,
>               aperfmperf,pni,dtes64,monitor,ds_cpl,vmx,smx,est,tm2,
>               ssse3,cx16,xtpr,pdcm,sse4_1,sse4_2,popcnt,lahf_lm,ida,
>               dts,tpr_shadow,vnmi,flexpriority,ept,vpid,model_Nehalem,
>               model_Conroe,model_coreduo,model_core2duo,model_Penryn,
>               model_n270
>    cpuModel = Intel(R) Xeon(R) CPU           X3430  @ 2.40GHz
>    cpuSockets = 1
>    cpuSpeed = 2393.769
> 
> 
> I compared libvirt's cpu_map.xml on both Centos 6.3 and CentOS 6.4 and
> indeed they do differ in large portions. So this patch should probably
> be merged to 3.1 branch? I will contact Dreyou and request that this
> patch will also be included in his builds. I guess otherwise there will
> be quite some fallout after people start picking CentOS 6.4 for oVirt 3.1.
> 
> Thanks again and best regards

Thank you for reporting this issue and verifying its fix.

I'm not completely sure that we should keep maintaining the ovirt-3.1
branch upstream - but a build destined for el6.4 must have it.

If you believe we should release a fix version for 3.1, please verify
that http://gerrit.ovirt.org/12723 has no ill effects.

Dan.