[Spacewalk-list] Mutli distro kickstart issue.

[Paul Robert Marino]

This is usually caused by errata synchronization tools.
There is a known bug in the API for spacewalk ERRATA publishing which
causes this. here is a link to a ticket which should give you a better
idea of the background
https://bugzilla.redhat.com/show_bug.cgi?id=834569

Essentially you have a channel cross contamination issue.
what you need to do is remove all of the offending packages except the
one written by the correct vendor from the channel.
Its not a problem with deduplication its a problem with erratas
spanning multiple base channels.

I have RHEL, Scientific Linux and CentOS on my spacewalk servers and
do not experience this issue.
this issue is one of the many reasons I wrote my own errata sync tool
here https://github.com/prmarino1/Extravehicular-Activity-Admin/tree/master/errata-tools
I ran into this same issue with rhn-clone-errata.py and the centos
version of the script which scrapes the centos mailing list as well.

Unfortunately the core API issue has never been fixed althouth the new
method of syncing erratas via the yum repo does not seem to be
effected by this so where possible try to use that method instead. the
problem is CentOS hasn't adopted this method and I'm not sure about
Oracle.
RHEL its pretty easy now to get the erratas using the new method if
you do a little research although I hesitate to write it out for you
completely.

In any case my script and a few others should be able to do eratta
syncs for you safely because 3 other developers and I all hit the same
issue when we wanted to create errata sync scripts which spanned
multiple base channels.

In fact my script can even sync erratas from one base channel to an
other on the same spacewalk server. so for CentOS in my test
environment I've synced the erratas from my Scientific Linux channel
to my CentOS channel. The down side is with syncing erratas from one
distro to an other using the method in my script you may have
incomplete erratas if the source distro ports a patch faster than the
destination distro, but its at least safe from this issue.




On Tue, Jul 1, 2014 at 11:39 AM, Canalichio, Kevin
<canalichio at jmail.metronaviation.com> wrote:
> I have set up my spacewalk server to have Redhat, Centos, and Oracle linux
> distros available for kickstart.
>
> When kickstarting I will get the following error:
>
>
>
>   “ The file libart_lgpl-2.3.20-5.1.el6.x86_64 cannot be opened. This is due
> to a missing file, a corrupted package or corrupted media.”
>
>
>
> The problem occurs because the md5 checksums are different for the same
> package name in the each of the different distros. I read somewhere that the
> satellite server uses some sort of dedup or links to not have multiple
> copies of the same file, which seems to work fine for updates, but not so
> well for kickstarts.
>
>
>
> I can fix this problem by clearing out all three the repos and resyncing the
> one I want to kickstart. But it take a while for the repo to rsync.
>
>
>
> I have also found this bug report that no one seems to be working on
> http://bugs.centos.org/view.php?id=6977 which references this from the
> archive
> https://www.redhat.com/archives/spacewalk-list/2014-February/msg00048.html
>
>
>
> Has anyone else seen this, or have a better work around?
>
>
> _______________________________________________
> Spacewalk-list mailing list
> Spacewalk-list at redhat.com
> https://www.redhat.com/mailman/listinfo/spacewalk-list