Re: [Xen-devel] [xen-4.2-testing test] 25126: regressions - FAIL

[Ian Campbell <Ian.Campbell@xxxxxxxxxx>]

On Wed, 2014-02-19 at 09:02 +0000, Jan Beulich wrote:
> >>> On 19.02.14 at 05:16, xen.org <ian.jackson@xxxxxxxxxxxxx> wrote:
> > flight 25126 xen-4.2-testing real [real]
> > http://www.chiark.greenend.org.uk/~xensrcts/logs/25126/ 
> > 
> > Regressions :-(
> > 
> > Tests which did not succeed and are blocking,
> > including tests which could not be run:
> >  test-amd64-i386-pv            9 guest-start               fail REGR. vs. 
> > 24865
> >  test-amd64-amd64-pv           9 guest-start               fail REGR. vs. 
> > 24865
> >  test-i386-i386-pv             9 guest-start               fail REGR. vs. 
> > 24865
> 
> These have been pretty persistent over the last couple of days,

Looks to me to be specific to xen-4.2-testing as well?

Previous successful run was against
640b31535ab8fe07911d0b90ae4adbe6078026c9

This is 6d83fd0a975731653efb0c470a4eeed6a44b99cd, the only thing in the
middle is "update Xen version to 4.2.4" which hardly seems likely to
have caused this.

> and this
> 
> Feb 18 13:36:04.556216   File "/usr/sbin/xend", line 36, in <module>
> Feb 18 13:36:04.564155     from xen.xend.server import SrvDaemon
> 
> looks very much like something broken in the infrastructure or the
> way how things get built.

I can see usr/lib/python2.7/site-packages/xen/xm and xend etc in the
dist.tar.gz built by the build job.

I wonder if we need to apply:
http://wiki.xen.org/wiki/Compiling_Xen_From_Source#Python_Prefix_and_Module_Layout

That might explain why only 4.2 (and I suppose earlier, if we were to
run such tests) are affected.

Ian, ts-xen-build has:

               ($ho->{Suite} =~ m/squeeze/ ? <<END : '').
        echo >>.config PYTHON_PREFIX_ARG=
END

I suspect this needs to actually be based upon the Xen version being
tested, not the host distro (or not only the host distro).

Newer versions of Xen (>=4.3) install to /usr/local by default and avoid
this issue, older versions of Xen need it on Wheezy as much as Squeeze. 

I don't think this workaround would be harmful on Wheezy+Xen4.3 but it
would be better to make it more targeted I think. That would imply a
runvar set by make-flight I think. What do you think?

Ian.


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