Partial mass rebuild for Python 2.7 coming soon (I hope)

David Malcolm dmalcolm at redhat.com
Thu Jul 22 15:28:21 UTC 2010


On Thu, 2010-07-22 at 08:39 -0400, Orcan Ogetbil wrote:
> On Thu, Jul 22, 2010 at 3:24 AM, Kevin Fenzi wrote:
> > On Thu, 22 Jul 2010 02:00:45 -0400
> > David Malcolm wrote:
> >
> > ...snip...
> >
> >> - build ordering was much more important than I hoped; most of the
> >> failures in this run seem to be due to incomplete deps in root.log.  I
> >> intend to retry these from the now-existing CVS tags, with a better
> >> build ordering (but I need to sleep first).
> >
> > Note that if you are not needing any changes in those packages you
> > could just resubmit the failed builds. So, perhaps list all the fails
> > and resubmit them all and see how many still fail?
> >
> 
> What if we need to make changes to make them build? Shall we wait
> until the mass rebuild is over or can we fix and build rightaway? If
> the latter, under which tag should we build?

Updated status:

The first set of builds are done:
  - 535 builds succeeded
  - 436 build failed
  - 1 build is stuck (python-cherrypy2-2.3.0-12.fc14.src.rpm, noarch
http://koji.fedoraproject.org/koji/taskinfo?taskID=2337424 )

Many of the build failures seem to be purely due to missing build
dependencies (because I didn't get ordering correct): you can see these
by the failure happening in the "root.log", rather than the "build.log"

I've kicked off a simple rebuild of everything that failed, using the
same CVS tag, into "dist-f14-py27-rebuild".  This will generate more
koji activity and notification spam (sorry), but shouldn't foul CVS and
changelogs in specfiles.

Hopefully this will shake out some more successful builds (python-nose
wasn't available in the buildroot for much of last night).

Once that's done, we can start looking at the more difficult failures.

Attached is the list of failures from the first build.  Having said
that, it's probably not particularly helpful yet, as I believe many of
these failed due to missing build deps.  You may want to wait until the
2nd attempt at rebuilding is done: I'll post an updated list of failures
then, and we can plan the next steps.

Having said that, some of the failures seem to be simple but real
issues, like extension modules missing a BuildRequires: python-devel

If you happen to see a genuine issue with one of your packages, feel
free to try and fix it: I've been doing rebuilds by hand on a few
high-importance packages.

You may have to wait until one of your BuildRequires is ready, of course
(or chase that one up).

The way I've been doing these manual rebuilds is to run this frmo a CVS
checkout of "devel":
  koji build dist-f14-py27-rebuild $(make cvsurl)
so that it rebuilds from the CVS tag into the correct
"dist-f14-py27-rebuild" target.   (Note that the normal target still has
python 2.6, not 2.7)

Q: How long does it take to populate the buildroot with this package?
A: It can take a while: as I understand it, once a build succeeds, a
"newRepo" koji task starts (unless one is already running) and that can
take a while.  Only when that's done does the buildroot have the new rpm
to build against.  As I understand it, there has to be a newRepo task
that started after the first build finished, and has finished, if that
makes sense

Hope this is helpful; sorry for the notification spam some of you are
getting from Koji.

Dave
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.fedoraproject.org/pipermail/devel/attachments/20100722/52604c76/attachment.html 


More information about the devel mailing list