[Spacewalk-list] Oddball question about Spacewalk possibly corrupting yum update

[P.Cookson at bham.ac.uk]

Hi Paul

I haven’t got any CentOS systems to patch, with Spacewalk, but a few thoughts…..

Not sure why you would start with 2.7 or sync from a local repo really? If you just didn’t want to go too bleeding edge with 2.9 I would at least start with 2.8. In addition, I wouldn’t bother syncing from a local repo as you can go direct to the CentOS web repo’s which are kept up to date – see eg’s below:

http://mirror.centos.org/centos/7/os/x86_64/                   #Base
http://mirror.centos.org/centos/7/extras/x86_64/           #Extras
http://mirror.centos.org/centos/7/updates/x86_64/       #Updates

You should find this combination will resolve any issues but, if not, you could sync the repo from the command line (using “reposync”) so you can watch it go through. However, there are repo sync logs in /var/log/rhn/reposync/ as well.

In addition, you could try manually updating a system from the command line (using “yum update”) and/or excluding  the X-Windows package you suspect (using “yum update -X <package>”), which may help you identify the problem too.

Unfortunately, Spacewalk does not provide an option to un-register a client system (similar to registering - “rhnreg_ks”) – the only option is to remove the client system’s profile from the Spacewalk server, then remove the /etc/sysconfig/rhn/systemid file etc on the client system.

Hope this is of help.

Regards
Phil





From: spacewalk-list-bounces at redhat.com <spacewalk-list-bounces at redhat.com> On Behalf Of paul.greene.va at gmail.com
Sent: 19 June 2019 17:00
To: spacewalk-list at redhat.com
Subject: [Spacewalk-list] Oddball question about Spacewalk possibly corrupting yum update

I built a Spacewalk 2.7 server for managing a couple hundred CentOS 6 workstations, and synced it to a local yum repository.

After a few months I started adding some CentOS 7 systems to the mix, and also synced to the same local yum repository (same server, different path to the repositories, of course).

The Spacewalk server always seemed to have an issue getting a successful full sync with the CentOS 7 local repository after the 7.6 series came out. I wasn't concerned at first because the systems were configured to go to the local yum repository anyway (configured in /etc/yum.repos.d).

After the CentOS 7.6 series came out, I started having an issue with any machine that got the 7.6 updates where the system would freeze and lock up, requiring a hard reboot to get it usable again. That happened on at least a daily basis and sometimes multiple times a day. I rolled those users back to CentOS 7.5 to get them a functioning stable machine again, and didn't update anybody that was running 7.5. It seemed definitely related to X windows - I could still go to a command prompt with ctrl-alt-f2 and work from a command prompt, but X windows wouldn't come back without a hard reboot. On a couple of servers that didn't need X windows and had the run level set to multi-user.target, they didn't have an issue - it was only the workstations that needed X windows.

I have access to another yum repository independent of the Spacewalk server, and noticed if I updated a workstation to that repository and did NOT join the machine to the Spacewalk server, they all ran fine with CentOS 7.6. As long as I didn't join them to the Spacewalk server there were no issues. I tried deleting the CentOS 7 yum repository from the Spacewalk server, thinking maybe the yum repository on the Spacewalk server had gotten corrupted and was pushing out some bad files, but that didn't work.

The systems still seem to be looking to the Spacewalk server for updates, regardless of what is in /etc/yum.repos.d.

Hopefully someone else has seen something like this before. I would either like to remove any and all yum configurations from the Spacewalk server, if possible, and just point to the local yum repository for any kind of updates.

Barring that, I'm interested in updating to Spacewalk 2.9 anyway, and would build a new 2.9 server, migrate everything over to that, and leave out any yum repository configuration at all on that new server.

The CentOS 6 systems are fine - no issues at all with updates and/or yum repositories.

Thanks,

PG
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/spacewalk-list/attachments/20190620/8eadf7c0/attachment.htm>