[Spacewalk-list] Issues upgrading to next point release

[Jan Pazdziora]

On Fri, Aug 03, 2012 at 09:06:26AM +0000, Ree, Jan-Albert van wrote:
> 
> > What happens if you schedule the update from Spacewalk's WebUI and
> > pick it up with rhn_check -vv -- will it get installed?
> 
> It tries but can't fetch the updates and gives a
> 
> D: Sending back response ((6,) 'Error while executing package action: empty transaction', {})

This I believe was a bug in rhn-client-tools (?) in RHEL 6 that
was fixed.

> > Spacewalk generates the repo data for the channels after any change
> > in the channel takes place, via an asynchronous job. The Last Repo
> > Build field in the channel page on the WebUI shows the timestamp of
> > the last repo generation, so you might want to check if it was updated
> > alright. I assume it was, otherwise your newly registered machine
> > wouldn't get them either, but we should know for sure.
> 
> Those dates are indeed very old... last build date is 10-7-2012 , last modified is 3-8-2012
> Looking at the task schedule, it shows "SKIPPED" since 10-7-2012
> So that looks like my main problem indeed... looking at the taskomatic daemon log I see Java heap space OutOfMemory errors...
> I've increased Xms and Xmx in /usr/share/rhn/config-defaults/rhn_taskomatic_daemon.conf and now it works again.

Good investigation. Not sure what version you Spacewalk server is but
we bumped up the taskomatic memory for 1.7.

> I have little over 10K packages in the channel that fails, which seems reasonable. Should I split things up further?

You shouldn't need to.

> > I have no idea what the sl-release package contains or overwrites,
> > so I won't speculate on this one. Does the package own the
> > /etc/sysconfig/rhn/systemid file? What does rpm -qf
> > /etc/sysconfig/rhn/systemid?
> 
> Systemid isn't owned by any package according to RPM
> 
> > Aren't some proxies between the clients and the Spacewalk server that
> > could be source of the problem?
> 
> No, the machines are on the same VLAN, with nothing but a switch between them.

OK.

If the taskomatic was failing and not regenerating the repo data,
I don't really see how the re-registered machine could have gotten the
content though. ;-)

-- 
Jan Pazdziora
Principal Software Engineer, Satellite Engineering, Red Hat