Trying to install rawhide: madness and where do we hope to go ?

[Daniel Veillard]

On Wed, Sep 07, 2005 at 08:15:38AM -0400, Jeff Spaleta wrote:
> On 9/7/05, Daniel Veillard <veillard at redhat.com> wrote:
> >   Can't basic bootup of kernels on i386 uniproc machine be tested before
> > pushed to rawhide, channels, mirrors and users ?
> http://www.livejournal.com/users/kernelslacker/23505.html
> 
> Are you going to provide the spare box to sit inside the RedHat fence
> line to watch for new kernel builds in redhat's buildsystem,test boot
> itself, and then report back in a scripted/automated fashion so that
> the kernel build can be removed?
> Until the binary build comes out in rawhide its not available for
> public consumption any other way from redhat's build system. So if you
> come up with a solution for this, it will have to be implemented
> completely within the redhat fenceline.

  Yes I agree with that last point.
Now when you say "Are you", if it's me Daniel Veillard, I don't think I'm
the best person to set this up. I'm certainly the best person to do the 
testing on the packages I maintain, I'm not sure me doing it for the kernel
will work best, and I probably would not use a new "box" but try to boot
the kernel/system though an emulator, maybe this just doesn't work, in which
case I would try to use a solution based on a new box and an hardware watchdog.
Maybe the consensus is that doing this ahead of push check is not worth it
but I would like to point out that there is a lot of human and physical 
resources affected in getting rawhide out and tested, and in comparison that
looks small to me.

> >   Similary if some change broke anaconda, can we backtrack that change
> > in the sake of keeping an installable dist, my servers are pulling everyday
> > new version of boot images and anaconda stuff which are know to not work,
> > and users are trying to use them while again we know they don't work.
> 
> Similarly are you going to provide the redhat internal hardware to do
> the scripted anaconda regression tests?

  In that case it's not a transcient failure, I think last time I tried to
use anaconda of rawhide back in July I had the exact same error. I think
everybody agree it is known to be broken, but hasn't been fixed or I was
just very very unlucky to get the same problem twice on 2 different hardware
at more than one month interval.

Daniel

-- 
Daniel Veillard      | Red Hat Desktop team http://redhat.com/
veillard at redhat.com  | libxml GNOME XML XSLT toolkit  http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/