On Wed, Mar 16, 2022 at 4:45 PM Demeter Tibor <tdemeter@itsmart.hu> wrote:
Dear Didi, Thank you for your reply.
My glusterfs uses the host's internal disks. I have 4 hosts, but the glusterfs use only 3. It is a centos7 based system.
As I think, I have to elminate the glusterfs first, because I can't upgrade the hosts until the engine running on there. That's why I think the engine will access to them after the reinstall.
Sorry, I do not think I understand your reasoning/plan. Please clarify.
Is it fine?
If the question is: "I have a HC hosted-engine+gluster on 3 hosts + another host. I want to move the hosted-engine (on one of the hosts? not clear) to external iscsi storage. Will the new engine be able to access the gluster storage on the existing hosts?" Then, sadly, I don't know. I *think* it will work - that gluster does not need the engine to function - but never tried this.
What pitfalls can I except in this case?
Many, if you ask me - mainly, because it's a flow probably hardly anyone ever tried, and it definitely goes out of the set of assumptions built into the basic design. I strongly suggest to test first on a test env (can be on nested-kvm VMs, no need for physical hosts) and to have good backups. But in theory, I can't think of any concrete point. Best regards,
Thanks in advance, Tibor
________________________________ Feladó: Yedidyah <didi@redhat.com> Címzett: Demeter <tdemeter@itsmart.hu> Másolat: users <users@ovirt.org> Dátum: 2022. március 16., szerda 15:08 CET Tantárgy: Re: [ovirt-users] Replace storage behind ovirt engine
On Wed, Mar 16, 2022 at 1:39 PM Demeter Tibor <tdemeter@itsmart.hu> wrote:
Dear Users,
I have to upgrade our hyperconverged ovirt system from 4.3 to 4.4, but meanwhile I would like to change the storage backend under the engine. At this moment it is a gluster based clustered fs, but I don't really like it.... I would like to change to a hardver based iscsi storage. I just wondering, when I do the engine reinstall, I will install it to an other storage. Is it possible somehow?
The "canonical" way is: 1. On current engine machine: engine-backup --file=f1 2. (Re)Install a new host (with 4.4?), copy there f1, and: hosted-engine --deploy --restore-from-file=f1
This only handles your engine VM, not others. The new engine (with its hosted_storage on the new iscsi storage) will need access to the existing gluster storage holding the other VMs, if you plan to somehow import/migrate/whatever them. If unsure how to continue, perhaps clarify your plan with more details.
Good luck, -- Didi
-- Didi