[Spacewalk-list] errata question

[Daniel Wittenberg]

Wouldn't this whole errata management be easier if you could point it at
a repo and have it one level higher, so it could see the OS and the
updates, and just mirror those over itself?  It seems like this is way
more complicated then it needs to be and I'm not understanding why it's
done the way it is.

Any insights?

Dan


On Thu, 2009-12-03 at 10:01 +0000, David Nutter wrote: 
> On Wed, Dec 02, 2009 at 06:14:16PM -0600, Daniel Wittenberg wrote:
> > Looking at this script I had a couple questions:
> > 
> > #All files for packages mentioned in the centos-announce postings 
> > #should reside in this directory. reposyncing the "updates" repo of
> > #your fave centos release should do the trick
> > package_dir=/centos-mirror/%(version)s.%(release)s/updates/%(arch)s/RPMS/
> > 
> > So say I have CentOS-Base channel, and I want all the errata to show up correctly in that channel.  It almost sounds like I need a 
> > different channel for the updates?
> 
> It's that way because I have a channel for each of the os, updates and
> extras CentOS upstream repositories. You can tell the script to push
> the errata to any channel though, it's the "update_channel" option
> later in the file. In your case you'd just push it to the base channel
> instead.
> 
> In the next version of the script (multiple arch support) I'll need to
> support many package_dirs anyway, so doing other configurations will
> become a bit easier too.
>  
> > Also, if I'm reposyncing the updates from the yum mirror, why do I
> > need a local copy as well?  Along those lines, if I'm pushing the
> > RPM's why would I also reposync them?
> 
> The package files are needed to get unambiguous NVREA values for each
> package. Parsing the package filename in the announce emails is
> error-prone[1] and there doesn't appear to be a way to get a package
> record from spacewalk using the package file name. Even if there was
> there might still be issues with (for example) noarch packages with
> the same filename in i386 and x86_64. If these are stored as separate
> package records in Spacewalk then you would have no way of knowing
> which was the "correct" one to associate with the errata.
> 
> Regards,
> 
> [1] That said I could try and hack something in to do this if people
> *really* want it.
>