Re: [Xen-devel] [xen-unstable test] 11831: regressions - FAIL

[Ian Jackson <Ian.Jackson@xxxxxxxxxxxxx>]

Jan Beulich writes ("Re: [Xen-devel] [xen-unstable test] 11831: regressions - 
FAIL"):
> >>> On 03.02.12 at 07:22, xen.org <ian.jackson@xxxxxxxxxxxxx> wrote:
> > Regressions which are regarded as allowable (not blocking):
> >  build-i386                    4 xen-build                fail   like 11637
> >  build-i386-oldkern            4 xen-build                fail   like 11637
> 
> Is that really the way things should work?

11637 broke because of a bug in the test harness relating to the new
qemu-upstream-unstable build.  I fixed that, but from the system's
point of view that means that 11637's builds weren't reliable either.

So this is a bug because we shouldn't be tolerating heisenbugs in the
build; any baseline passed build should be enough to block a push if
the corresponding staging build failed.  I've made what I think is a
suitable change, which is now undergoing testing to check that it
works.

There's also a separate bug which is that it failed to notice for many
tests that they were blocked by lack of the build:

> >  test-amd64-i386-xl            4 xen-install          fail REGR. vs. 11643

But happily this blocked the push of the broken build.  I will push a
fix for this other bug when the first bugfix has made it through the
tester.

Ian.

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel