Re: [ovirt-users] Upgrade 4.0.6 to 4.1.1 -- How to Update oVirt Node (4.x) Hosts?

Didi, Thanks for the tip on the utilities – I’ll add that for future upgrades. Since you pointed that out, I’m reminded that in a previous upgrade (following one of the developer’s suggestions) I had added this: /etc/ovirt-engine/engine.conf.d/99-custom-truststore.conf So I guess that’s why my https certificate was preserved.

As to the documentation, I did submit a pull request (#923) and ‘JohnMarksRH’ added that along with some additional edits. I’ll move any continuing discussion on that to another thread. And yes, the RHV documentation is excellent and I’ve often turned to it. It’s too bad some of the effort ends up being duplicated. Anyway…. Here’s what I did with one of the oVirt nodes: yum -y remove ovirt-release40 yum -y install http://resources.ovirt.org/pub/yum-repo/ovirt-release41.rpm cd /etc/yum.repos.d # ls CentOS-Base.repo CentOS-fasttrack.repo CentOS-Sources.repo cockpit-preview-epel-7.repo CentOS-CR.repo CentOS-fasttrack.repo.rpmnew CentOS-Vault.repo ovirt-4.0-dependencies.repo CentOS-Debuginfo.repo CentOS-Media.repo CentOS-Vault.repo.rpmnew ovirt-4.0.repo rm -f ovirt-4.0* After doing that, when I check again in the admin GUI for an upgrade, it shows one available (4.1.1.1). From the GUI I tell it to upgrade, and it runs along with no errors, seems to finish, and then reboots the host. When the host comes back up, it’s still running 4.0.6. When I check again for an available upgrade, it doesn’t see it available. I’m attaching the installation log that is referenced in Events in the GUI. If I go straight into the node and run ‘yum update’ and reboot, then it gets the latest 4.1.x image and the engine detects it as such.

But of course that’s not the ideal method. I used the manual method for the remaining hosts. I don’t know if this is related, but since the upgrade I’ve also noticed an unfamiliar error when I log in directly to the engine host. (It’s a standalone Centos7 VM running on a separate KVM host.) Here is is: nodectl must be run as root! nodectl must be run as root! This comes up when *any* user logs into the box. When I switch to root I get this: /bin/python3: Error while finding spec for 'nodectl.__main__' (<class 'ImportError'>: No module named 'nodectl') /bin/python3: Error while finding spec for 'nodectl.__main__' (<class 'ImportError'>: No module named 'nodectl') So it looks like it’s been invoked from here: ls -llh /etc/profile.d/nodectl* -rwxr-xr-x. 1 root root 13 Apr 6 06:46 /etc/profile.d/nodectl-motd.sh -rwxr-xr-x. 1 root root 24 Apr 6 06:46 /etc/profile.d/nodectl-run-banner.sh According to ‘yum whatprovides’ this appears to have been installed by package “ovirt-node-ng-nodectl-4.1.0-0.20170406.0.el7.noarch”. Anyone else getting this? I can try fixing the python error by adding the module, but I thought I’d report this first. Any suggestions as to next steps?

Thanks Daniel On 4/25/17, 2:01 AM, "Yedidyah Bar David" <didi(a)redhat.com&gt; wrote: On Tue, Apr 25, 2017 at 1:19 AM, Beckman, Daniel <Daniel.Beckman(a)ingramcontent.com&gt; wrote: > So I successfully upgraded my engine from 4.06 to 4.1.1 with no major > issues. > > > > A nice thing I noticed was that my custom CA certificate for https on the > admin and user portals wasn’t clobbered by setup. > > > > I did have to restore my custom settings for ISO uploader, log collector, > and websocket proxy: > > cp > /etc/ovirt-engine/isouploader.conf.d/10-engine-setup.conf.<latest_timestamp> > /etc/ovirt-engine/isouploader.conf.d/10-engine-setup.conf > > cp > /etc/ovirt-engine/ovirt-websocket-proxy.conf.d/10-setup.conf.<latest_timestamp> > /etc/ovirt-engine/ovirt-websocket-proxy.conf.d/10-setup.conf > > cp > /etc/ovirt-engine/logcollector.conf.d/10-engine-setup.conf.<latest_timestamp> > /etc/ovirt-engine/logcollector.conf.d/10-engine-setup.conf The utilities read these files sorted by name, last wins. So you can add '99-my.conf' to each and have it override whatever engine-setup does. > > > > Now I’m moving on to updating the oVirt node hosts, which are currently at > oVirt Node 4.0.6.1. (I’m assuming I should do that before attempting to > upgrade the cluster and data center compatibility level to 4.1.) > > > > When I right-click on a host and go to Installation / Check for Upgrade, the > results are ‘no updates found.’ When I log into that host directly, I notice > it’s still got the oVirt 4.0 repo, not 4.1. Is there an extra step I’m > missing? The documentation I’ve found > (http://www.ovirt.org/documentation/upgrade-guide/chap-Updates_between_Min...) > doesn’t mention this. You are right. It's mentioned for the engine in the release notes [1] but not for the hosts. Please file a github issue or send a pull request :-) [1] https://www.ovirt.org/release/4.1.0/ > > > > > > ** > > If I can offer some unsolicited feedback: I feel like this list is populated > with a lot of questions that could be averted with a little care and feeding > of the documentation. It’s unfortunate because that makes for a rocky > introduction to oVirt, and it makes it look like a neglected project, which > I know is not the case. Patches are welcome :-) > > > > On a related note, I know this has been discussed before but… > > The centralized control in Github for the documentation does not really > encourage user contributions. What’s wrong with a wiki? If we’re really > concerned about bad or malicious edits being posted, keep the official in > git and add a separate wiki that is clearly marked as user-contributed. That was indeed discussed in the past, I am not aware of any conclusions. Perhaps start a separate thread about this? Adding Duck. Please also note that you can have a look at RHV documentation [2]. Almost all of it applies to oVirt as well (and oVirt's to RHV). [2] https://access.redhat.com/documentation/en/red-hat-virtualization/ Best, > > ** > > > > > > Thanks, > > Daniel > > > _______________________________________________ > Users mailing list > Users(a)ovirt.org > http://lists.ovirt.org/mailman/listinfo/users > -- Didi