Re: [xen-4.5-testing test] 58276: regressions - FAIL

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

On Wed, 2015-06-10 at 10:48 +0100, Jan Beulich wrote:
> >>> On 10.06.15 at 05:28, <osstest@xenbits.xen.org> wrote:
> > flight 58276 xen-4.5-testing real [real]
> > http://logs.test-lab.xenproject.org/osstest/logs/58276/ 
> > 
> > Regressions :-(
> > 
> > Tests which did not succeed and are blocking,
> > including tests which could not be run:
> >  test-amd64-i386-xl-qemuu-winxpsp3 15 guest-localmigrate/x10 fail REGR. vs. 57908
> 
> I just looked at this and ...
> 
> > Regressions which are regarded as allowable (not blocking):
> >  test-armhf-armhf-xl-sedf      6 xen-boot                  fail REGR. vs. 57908
> >  test-amd64-i386-xl-qemut-win7-amd64 16 guest-stop              fail like 57908
> >  test-amd64-amd64-xl-qemuu-win7-amd64 15 guest-localmigrate/x10 fail like 57908
> 
> ... this another time: The former is on Intel, while the latter on AMD,
> so unlikely (i.e. minus one of them being some random, unrelated
> failure) to be host or even vendor specific. Since this test consists
> of 10 migrations in a row, with guest liveliness only checked once at
> the end, it's rather unlikely that we'll be able to tell anything from
> looking at the logs taken at the end. Yet the guest not responding
> to pings first of all raises the question of whether its interrupts are
> somehow not arriving anymore as intended. Would it be possible to
> look at such a guest in that final state to check whether e.g.
> keyboard input / mouse movement still work? Or to take a series of
> xenctx or xl vcpu-list outputs to see whether the guest is still doing
> _something_ (i.e. not completely locked up)?

Osstest doesn't (AFAIK) have a mechanism for deciding to preserve a test
machine upon failure for later fiddling with.

The only choice would be to manually lock the machine and run adhoc
reproduction until the issue occurred.

> 
> Also, has anyone with more tools/qemu/osstest knowledge than me
> managed to gain at least some weak theory of where there might be
> this qemuu dependency affecting _only_ the stable trees?
> 
> Jan
> 
> 
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel