Re: [xen-4.5-testing baseline-only test] 38676: trouble: blocked/broken/fail/pass

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

On Thu, 2016-01-21 at 07:49 +0100, Juergen Gross wrote:
> On 21/01/16 07:31, Platform Team regression test user wrote:
> > This run is configured for baseline tests only.
> > 
> > flight 38676 xen-4.5-testing real [real]
> > http://osstest.xs.citrite.net/~osstest/testlogs/logs/38676/
> 
> I've seen this more than once now: some test results seem not to be
> reachable from outside Citrix:
> 
> $ host osstest.xs.citrite.net
> Host osstest.xs.citrite.net not found: 3(NXDOMAIN)
> 
> Is this on purpose?

More due to a lack of somewhere to post them publicly.

These results from "citrix-osstest@xenproject.org", generally with
"baseline-only" in the subject and a flight > 5000 are from the osstest
instance running in Cambridge, which used to be the production instance
prior to the colo.

Reports from the production colo come from "osstest service owner <osstest-
admin@xenproject.org>".

The Cambridge baseline-only tests were originally setup so we could make
use of the historical info e.g. regarding machine specific failures in the
period after the switch (i.e. when there wasn't much historical data for
the colo based machines) and to give some overlap in the results of the old
and new instances.

Since then I have left them running since we have versions of hardware
(particularly older machines) which the COLO doesn't have, and also because
it gives those of us here a set of baselines to use for any adhoc testing
we might run here.

Being "baseline-only" each branch is run exactly once on any new baseline,
i.e. when the XenProject instance gets a push of that branch.

Right now our instance is suffering from a host-install issue exposed by
the Debian Jessie upgrade (arising because some of those machines have two
disks, which caused us to notice that none of the machines in the COLO did,
which is being fixed), so the results look bad right now. The fix is in
transit through the colo push gate where it will be picked up by the
Cambridge instance and the results should improve.

Anyway, I kept the mails going to the list because they might be of
interest, and because they are baseline-only they should be relatively low
volume (compared with the production instance). I'm happy to push logs for
specific flights to my user space on xenbits if someone wants to look at an
issue, but for the most part I would expect them to be ignored. If they are
considered too annoying I can arrange for the reports to go elsewhere.

Ian.