Re: Xen 4.10 breakage with buster (was Re: [xen-4.10-testing test] 151033: regressions - trouble: blocked/fail/pass/starved) [and 1 more messages]

[Ian Jackson <ian.jackson@xxxxxxxxxx>]

Jan Beulich writes ("Re: Xen 4.10 breakage with buster (was Re: 
[xen-4.10-testing test] 151033: regressions - trouble: 
blocked/fail/pass/starved) [and 1 more messages]"):
> Right - which will then enable 4.10's -prev build to work, which
> will in turn allow 4.11's -prev builds to work, and then the same
> for 4.12. I.e. osstest will continue to be quite busy for the
> next several days.

Yes.

> Instead of building -prev anew each time, wouldn't it make sense
> to store and re-use the most recent "prev" tree's build? This
> ought to avoid this sort of cascade dependencies in particular
> when upgrading the test platform underneath. There's no reason
> to fail a flight just because the N-1 tree doesn't build anymore.

Well.  In principle, yes.  Fairly recently osstest got machinery to do
that kind of anointed build outputs.  It's not entirely trivial, and
this kind of thing happens about once every two years.  We're noticing
two such cases in a row because we put off the osstest stretch upgrade
and are now doing the buster one early.

Ian.