Re: stable trees (was: [xen-4.2-testing test] 58584: regressions)

[Ian Campbell <ian.campbell@citrix.com>]

On Fri, 2015-06-19 at 12:07 +0100, Ian Campbell wrote:
> On Fri, 2015-06-19 at 10:51 +0100, Jan Beulich wrote:
> > >>> On 18.06.15 at 16:22, <ian.campbell@citrix.com> wrote:
> > > On Thu, 2015-06-18 at 12:37 +0100, Jan Beulich wrote:
> > >> >>> On 17.06.15 at 12:26, <Ian.Jackson@eu.citrix.com> wrote:
> > >> > Jan Beulich writes ("stable trees (was: [xen-4.2-testing test] 58584: 
> > >> > regressions)"):
> > >> >> Which leaves several options:
> > >> >> - the problem was always there, but hidden by some factor in the
> > >> >>   old osstest instance,
> > >> > 
> > >> > I think this is most likely.  The old system had much older hosts.
> > >> > 
> > >> > I think this is a race that we now happen to lose most of the time.
> > >> 
> > >> For verification purposes, would it be possible to set up a couple of
> > >> flights on the old instance for one of the stable trees?
> > > 
> > > I can try and run something adhoc on the old system if you can let me
> > > know exactly which jobs (test-*-*-*) and branches you are interested in.
> > 
> > Any or all of test-amd64-*-xl-qemuu-win* (not sure whether you
> > can specify wildcards), and I guess stable-4.5 (or staging-4.5)
> > would be the most natural branch choice.
> 
> I think the tools can do wildcards, yes.
> 
> I've kicked off a full adhoc xen-4.5-testing flight so I have a local
> template to copy the jobs from for some repeated runs with just the
> problem flights (it's just easier to do that than to invent a cut-down
> flight from scratch...).

After that baseline I ran a few tests of just the windows + qemuu stuff:
http://xenbits.xen.org/people/ianc/tmp/adhoc/37619/

was allowing free reign on the machines and was mostly successful, apart
from the windows-install failure on lake-frog. Looking at the test
history this seems to have always been a problem on the old infra.
*-frog are "AMD Opteron(tm) Processor 6168" which is as close as the old
infra has to the new colos merlot[01] which is "AMD Opteron(tm)
Processor 6376".

With that in mind I reran with things limited to the two frog-* boxes
and got http://xenbits.xen.org/people/ianc/tmp/adhoc/37624/.

The windows-install of winxpsp3 persisted but there was no migration
failure elsewhere.

It's not a lot of data, but in comparison with the results in the colo:
http://logs.test-lab.xenproject.org/osstest/results/history/test-amd64-amd64-xl-qemuu-win7-amd64/xen-4.5-testing.html 
it looks like it's the newer system which is exposing the issue.

Ian.