10:15 a.m.
On Tue, Dec 3, 2019 at 3:44 PM Robert Webb <rwebb(a)ropeguru.com> wrote:
Didi,
Thanks for the reply.
Yes, backups are done, I am just thinking in the case of Murphy's Law where nothing
works and the engine is gone. I am also new to oVirt and am always loking for the
"proper" backup method as you have laready mentioned using engine-backup.
Storage is NFS backed and backups are done in case of storage server failure.
OK.
I suggest, then, that you methodically plan, document and test your
environment on various different failure modes.
Since you are new to oVirt, it's obvious you still do not have the
full mental image of all the components, their behaviour, etc., and
even more so how they might react, or be handled, in emergency situations.
In particular, spend some time learning about the SPM, master storage
domain, HA VMs and how they might be affected by/affecting restore flows,
etc. Good luck!
What is the preferred VM backup method so both disk and VM config is captured? Would
rather not have to put agents on all the VM for OS level backups and then have to manually
recreate all of the machines. Do all the VM configs get stored on the shared storage just
like one of the other virtualization vendors?
oVirt does not supply, nor endorse, a specific "preferred method" for this.
Rather, it exposes an API, and provides a simple example script:
https://github.com/oVirt/ovirt-engine-sdk/blob/master/sdk/examples/vm_bac...
There are several backup products/projects, both large and small, both
open and closed source, that use this API and can be used. I personally
never used any, so can't recommend anything. If you search a bit the list
archives you'll find at least some of them.
Best regards,
Appreciate the time from folks on the list who respond.
________________________________
From: Yedidyah Bar David <didi(a)redhat.com>
Sent: Tuesday, December 3, 2019 3:29 AM
To: Robert Webb <rwebb(a)ropeguru.com>
Cc: users <users(a)ovirt.org>
Subject: Re: [ovirt-users] Scenario: Ovirt-engine on hardware and recovery
On Tue, Dec 3, 2019 at 4:38 AM <rwebb(a)ropeguru.com> wrote:
>
> So thinking about my setup, I am thinking through different failure scenarios.
>
> So lets say I have a small physical server with 8 cores and 16GB RAM and I install
Centos 7 and ovirt-engine on bare metal. This would also be the same scenario is the
engine were on a VM.
>
> I run into a major hardware issue and completely lose the engine.
You do take backups, right? :-)
For the engine itself, you can use engine-backup.
>
> How does one recover the cluster setup and not have to start from scratch by having
to rebuild all the nodes? Can the engine just be rebuilt and the ovirt nodes be imported?
You didn't mention your plans re storage.
If you put your VMs on well-handled storage, and lose everything but
the storage, you can install a new engine, new nodes, and then import
the VMs from the existing storage. You will still miss some
configuration etc., but should mostly be ok.
If you plan to use gluster storage using the same nodes, aka HCI, the
same applies, obviously, only that the nodes are your storage, and not
an external system. You still have to make sure you handle them
properly - backup, DR, etc., depending on your needs and budget.
If you do have engine-backup backups, but lost all your storage,
engine-backup will not help you.
>
> This scenario is based on the nodes being built from the ovirt node iso.
That's ok, but is not relevant. Main question is the storage. Can be
external or local/gluster with either ovirt-node or EL.
Good luck,
--
Didi
--
Didi