Re: [Xen-devel] [xen-unstable test] 57852: regressions - FAIL

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 10.06.15 at 10:50, <ian.campbell@xxxxxxxxxx> wrote:
> On Tue, 2015-06-09 at 10:29 +0100, Jan Beulich wrote:
>> >>> On 09.06.15 at 10:26, <ian.campbell@xxxxxxxxxx> wrote:
>> > On Mon, 2015-06-08 at 13:16 +0100, Ian Campbell wrote:
>> > 
>> >> The adhoc run passed, but that's not statistically significant.
>> > 
>> > I ran a bunch more in this no-apicv configuration, the logs are at
>> > http://logs.test-lab.xenproject.org/osstest/logs/<NNNN>:
>> > 
>> > Flight     Host    Failed at
>> > 58190      fiano0  ts-guest-stop
>> > 58198      fiano0  ts-guest-stop
>> > 58203      fiano0  ts-windows-install
>> > 58208      fiano0  ts-guest-stop
>> > 58210      fiano1  ts-guest-stop
>> > 58214      fiano1  ts-guest-stop
>> > 58217      fiano1  ts-guest-stop
>> > 
>> > I think that's not sufficient data to draw a conclusion, since there was
>> > always a small background failure rate. I'm going to run another half
>> > dozen (3 on each).
>> 
>> At least the one failure is following the patterns of previous ones
>> (ping timing out and guest sitting with both of its vCPU-s on HLT,
>> and the last VM entry having delivered a timer interrupt). Without
>> knowing _when_ that last timer interrupt got injected and whether
>> other interrupts are occurring for the guest as necessary, that
>> again doesn't mean much.
> 
> 58243 fiano0  ts-guest-stop
> 58251 fiano0  ts-guest-stop
> 58258 fiano0  ts-guest-stop
> 58266 fiano1  ts-guest-stop
> 58279 fiano1  ts-guest-stop
> 58282 fiano1  ts-guest-stop
> 
>> > I'm also going to run some on another pair of hosts without no-apicv. I
>> > chose elbling[01] since according to
>> > 
> http://logs.test-lab.xenproject.org/osstest/results/history.test-amd64-amd64 
>> > -xl-qemuu-win7-amd64.html it has been used a few times on various branches 
>> > and hasn't so far failed a ts-windows-install. It looks to have the same 
> set 
>> > of advanced features as fiano*:
>> 
>> That's a good idea, thanks.
> 
> 58244 elbling0        ts-guest-stop
> 58250 elbling0        ts-guest-stop
> 58256 elbling0        ts-guest-stop
> 58261 elbling1        ts-guest-stop
> 58269 elbling1        ts-guest-stop
> 58274 elbling1        ts-guest-stop
> 
> So it is looking awfully like a host specific issue with apicv on
> fiano*.

Indeed. Leaving us with the slight hope that there is a microcode
update available that's newer than what the BIOS of those boxes
loads. Could we perhaps afford un-blessing the two systems for
the time being? And maybe get Intel involved if there's no ucode
update available that helps?

Jan


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel