Thank you for the explanation. My misunderstanding, then. It seems that I could get away with only including centos-release-ovirt41 in the past, but not any more. Once I did that I was *almost* able to upgrade some of the hosts (I will do the rest later). Warnings for others: I had a custom /etc/multipath.conf , since it started with the line "# VDSM REVISION 1.3" , vdsm refused to start (There was a clear warning of this in /var/log/messages). Once I changed the first 2 lines to: # VDSM REVISION 1.4 # VDSM PRIVATE Everything was fine. I cannot update userspace-rcu.x86_64 from 0.7 to 0.10 because of this bug. https://bugzilla.redhat.com/show_bug.cgi?id=1507090  No big deal for me since I am not using gluster, but I guess that basically everybody will be getting the same. All my hosts show a warning in the general tab in the UI with an exclamation icon and the text "Host has no default route". I indeed have a valid default gateway on the management interface for all my compute hosts. I like the new UI. Thanks for your help! Eduardo Mayoral Jimeno (emayoral@arsys.es) Administrador de sistemas. Departamento de Plataformas. Arsys internet. +34 941 620 145 ext. 5153 On 28/12/17 13:23, Yedidyah Bar David wrote:

On Thu, Dec 28, 2017 at 1:46 PM, Eduardo Mayoral <emayoral@arsys.es> wrote:

...

Hi,

I just started upgrading one of our oVirt installations from 4.1.8 to 4.2

Engine is NOT self hosted. It is a plain CentOS 7.4 server.

Compute hosts (10 of them) are CentOS 7.4 as well. Storage is mainly iSCSI served from an external storage system.

Engine upgrade was seamless. Now I want to upgrade the hosts, then change the cluster compatibility level, then the datacenter compatibility. Can someone confirm that this is the correct sequence?

On the first host, I put it in maintenance mode, installed centos-release-ovirt42 and removed centos-release-ovirt41.

When doing the upgrade host from the web interface, I get:

2017-12-28 10:58:46,054 p=7173 u=ovirt | Using /usr/share/ovirt-engine/playbooks/ansible.cfg as config file 2017-12-28 10:58:46,288 p=7173 u=ovirt | PLAY [all] ********************************************************************* 2017-12-28 10:58:46,316 p=7173 u=ovirt | TASK [ovirt-host-upgrade : Install ovirt-host package if it isn't installed] *** 2017-12-28 10:58:51,143 p=7173 u=ovirt | fatal: [llki612.arsyslan.es]: FAILED! => {"changed": false, "msg": "No package matching 'ovirt-host' found available, installed or updated", "rc": 126, "results": ["No package matching 'ovirt-host' found available, installed or updated"]} 2017-12-28 10:58:51,144 p=7173 u=ovirt | PLAY RECAP ********************************************************************* 2017-12-28 10:58:51,144 p=7173 u=ovirt | llki612.arsyslan.es : ok=0 changed=0 unreachable=0 failed=1

Now, I see the ovirt-host package is not installed or available on the Compute node. It is available on the ovirt-engine on repo "ovirt-4.2" (which is installed by ovirt-release42-4.2.0-1.el7.centos.noarch)

I am a bit confused here. I thought I only needed centos-release-ovirt42 (centos-ovirt42 repo) on the compute hosts. Is that so? Has this changed from 4.1 to 4.2? I do not think this was changed.

Where did you find instructions to install centos-release-ovirt42?

Generally speaking, oVirt repos are available by installing ovirt-release*.

Best regards,

...

Thank you very much for your help!

-- Eduardo Mayoral Jimeno (emayoral@arsys.es) Administrador de sistemas. Departamento de Plataformas. Arsys internet. +34 941 620 145 ext. 5153

_______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users