the better way to update vdsm

This is a multi-part message in MIME format. --------------020207020907080505090303 Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 7bit Hi all, I'm about to update to 3.4.4 with the associated vdsm. Updating vdsm with yum is trivial, but it usually breaks running vms on the host. So these are the steps I follow each time: * live migrating runnings vms to a another host in the cluster * putting the host in maintenance * updating vdsm * activating the host * getting back the vms * and so on with the other hosts... But sometimes it is not enough, I need to reinstall the host so as to the host be able to successfully activate. When getting back the vms on the initial host, I must do it manually to load balance vms between the hosts and this is not very convinient. So my question is : do I update in the right way, or is there any recommendations that I didn't find anyway? Thanks. --------------020207020907080505090303 Content-Type: text/html; charset=utf-8 Content-Transfer-Encoding: 7bit <html> <head> <meta http-equiv="content-type" content="text/html; charset=utf-8"> </head> <body bgcolor="#FFFFFF" text="#000000"> Hi all,<br> <br> I'm about to update to 3.4.4 with the associated vdsm. Updating vdsm with yum is trivial, but it usually breaks running vms on the host.<br> So these are the steps I follow each time:<br> <ul> <li>live migrating runnings vms to a another host in the cluster</li> <li>putting the host in maintenance</li> <li>updating vdsm</li> <li>activating the host</li> <li>getting back the vms</li> <li>and so on with the other hosts...<br> </li> </ul> <p>But sometimes it is not enough, I need to reinstall the host so as to the host be able to successfully activate.<br> </p> <p>When getting back the vms on the initial host, I must do it manually to load balance vms between the hosts and this is not very convinient.<br> </p> <p>So my question is : do I update in the right way, or is there any recommendations that I didn't find anyway?<br> </p> <p>Thanks.<br> </p> </body> </html> --------------020207020907080505090303--

Hello Nathanaël, I think the supposed way to upgrade would be: * (hosted-engine only) put cluster in global maintenance if running hosted-enfgine * upgrade Engine using engine-setup (this *should* detect product updates) * (hosted-engine only) disable global maintenance * Put host in maintenance mode (witch will migrate VMs automatically) * Click 'reinstall host' in Engine (this step will will invoke yum update?) * Activate host continue with the other hosts If there are kernel updates, I usually update them as well in the process (involving manual yum). Can anyone clarify if 'reinstall host' invokes yum? Thanks! On 24.09.2014 11:41, Nathanaël Blanchet wrote:
Hi all,
I'm about to update to 3.4.4 with the associated vdsm. Updating vdsm with yum is trivial, but it usually breaks running vms on the host. So these are the steps I follow each time:
* live migrating runnings vms to a another host in the cluster * putting the host in maintenance * updating vdsm * activating the host * getting back the vms * and so on with the other hosts...
But sometimes it is not enough, I need to reinstall the host so as to the host be able to successfully activate.
When getting back the vms on the initial host, I must do it manually to load balance vms between the hosts and this is not very convinient.
So my question is : do I update in the right way, or is there any recommendations that I didn't find anyway?
Thanks.
-- Daniel Helgenberger m box bewegtbild GmbH P: +49/30/2408781-22 F: +49/30/2408781-10 ACKERSTR. 19 D-10115 BERLIN www.m-box.de www.monkeymen.tv Geschäftsführer: Martin Retschitzegger / Michaela Göllner Handeslregister: Amtsgericht Charlottenburg / HRB 112767

Hello Nathanaël,
I think the supposed way to upgrade would be: Here you need to insert: # yum update engine-setup on the engine (and for a new major version also new .repo files)
Replying to my own mail since upgrading opportunities are not many. Considering the procdure mentioned worked very well, it needs some refinement. On 24.09.2014 13:30, Daniel Helgenberger wrote: this is necessary since the engine-setup utility cannot upgrade itself (though this might be doable; RFE? :)
* (hosted-engine only) put cluster in global maintenance if running hosted-enfgine * upgrade Engine using engine-setup (this *should* detect product updates) * (hosted-engine only) disable global maintenance * Put host in maintenance mode (witch will migrate VMs automatically)
This did not work for the hosted engine VM for some reason; I migrated it manually. Is this intentional?
* Click 'reinstall host' in Engine (this step will will invoke yum update?) * Activate host
continue with the other hosts
If there are kernel updates, I usually update them as well in the process (involving manual yum).
Can anyone clarify if 'reinstall host' invokes yum? Thanks! It will run yum update; but only on oVirt packages. One might want to run it again to update all other system packages. Again, this makes sense.
On 24.09.2014 11:41, Nathanaël Blanchet wrote:
Hi all,
I'm about to update to 3.4.4 with the associated vdsm. Updating vdsm with yum is trivial, but it usually breaks running vms on the host. So these are the steps I follow each time:
* live migrating runnings vms to a another host in the cluster * putting the host in maintenance * updating vdsm * activating the host * getting back the vms * and so on with the other hosts...
But sometimes it is not enough, I need to reinstall the host so as to the host be able to successfully activate.
When getting back the vms on the initial host, I must do it manually to load balance vms between the hosts and this is not very convinient.
So my question is : do I update in the right way, or is there any recommendations that I didn't find anyway?
Thanks.
-- Daniel Helgenberger m box bewegtbild GmbH P: +49/30/2408781-22 F: +49/30/2408781-10 ACKERSTR. 19 D-10115 BERLIN www.m-box.de www.monkeymen.tv Geschäftsführer: Martin Retschitzegger / Michaela Göllner Handeslregister: Amtsgericht Charlottenburg / HRB 112767

On Wed, Sep 24, 2014 at 11:30:03AM +0000, Daniel Helgenberger wrote:
Hello Nathanaël,
I think the supposed way to upgrade would be:
* (hosted-engine only) put cluster in global maintenance if running hosted-enfgine * upgrade Engine using engine-setup (this *should* detect product updates) * (hosted-engine only) disable global maintenance * Put host in maintenance mode (witch will migrate VMs automatically) * Click 'reinstall host' in Engine (this step will will invoke yum update?) * Activate host
continue with the other hosts
If there are kernel updates, I usually update them as well in the process (involving manual yum).
Can anyone clarify if 'reinstall host' invokes yum? Thanks!
On 24.09.2014 11:41, Nathanaël Blanchet wrote:
Hi all,
I'm about to update to 3.4.4 with the associated vdsm. Updating vdsm with yum is trivial, but it usually breaks running vms on the host. So these are the steps I follow each time:
* live migrating runnings vms to a another host in the cluster * putting the host in maintenance * updating vdsm * activating the host
That's the prudent thing to do, but `yum upgrade vdsm` should NOT "break vms". When it does, I'd love to see a bug report (VMs being temporarily paused due to lost storage is expected).
participants (3)
-
Dan Kenigsberg
-
Daniel Helgenberger
-
Nathanaël Blanchet