On Tue, Jan 25, 2022 at 8:57 PM Nir Soffer <nsoffer(a)redhat.com> wrote:
On Tue, Jan 25, 2022 at 7:21 PM Richard W.M. Jones <rjones(a)redhat.com> wrote:
>
> On Tue, Jan 25, 2022 at 06:37:24PM +0200, Nir Soffer wrote:
> > On Tue, Jan 25, 2022 at 5:46 PM Richard W.M. Jones <rjones(a)redhat.com>
wrote:
> > >
> > > A while back I had oVirt 4.4.7 installed which I used for testing.
> > > For some reason that installation has died in some way, so I'm trying
> > > to install a fresh new oVirt 4.4.10.
> > >
> > > Last time I installed ovirt, it was very easy - I provisioned a couple
> > > of machines, ran engine-setup in one, answered a few questions and
> > > after a few minutes the engine was installed.
This is still the situation, if you install a "standalone" engine:
https://www.ovirt.org/documentation/installing_ovirt_as_a_standalone_mana...
> > >
> > > Somehow this has changed and now it's really far more complicated,
> > > involving some ansible things and wanting to create VMs and ssh
> > > everywhere.
Perhaps you tried a self-hosted engine installation?
https://www.ovirt.org/documentation/installing_ovirt_as_a_self-hosted_eng...
> > >
> > > Can I go back to the old/easy way of installing oVirt engine? And if
> > > so, what happened to the instructions for that?
> >
> > engine-setup still works, maybe you can give move details on what went wrong?
>
> So I managed to dnf install /usr/bin/engine-setup. When I ran it, it
> wanted to connect to an external PostgreSQL server. I'm pretty sure
> that never happened last time.
Was that a clean machine, with no left-overs?
Do you see the connection attempts interactively, or only if you check
the setup log?
>
> TBH I'm also going to erase everything and start again because I've
> been round several loops here already.
If the machine has cruft accumulated, this is often the simplest approach,
if you do not need the data there.
>
> Do we recommend RHEL or CentOS as the initial OS?
For 4.4.10 your best option is RHEL 8.5.
At least in theory. Do you do this yourself? Heard of others doing it?
I personally only use, for development, either CentOS Stream 8 + oVirt,
or RHEL+RHV. Over the years we did get reports of people using RHEL+oVirt,
and also fixed a few bugs around this. But if not many people do, we'll
not hear about the problems, if any, so won't fix them...
I'm not sure about the status of the RHEL clones (e.g. Rocky). In the past
We did get here a few reports about this. I didn't try this myself.
Generally speaking, it should work, with current master (4.5), and if
not, should be rather easy to patch/fix. Not sure about 4.4.
they were missing the advanced virtualization packages, and could be
installed using Centos advanced virtuatation packages which probably are not
available now (EOL).
Not sure why you think so. Isn't that exactly the same thing people
are using when using CentOS? What's the difference? Indeed, if people
want to use Rocky/Alma/etc and insist on not using anything from CentOS,
they'll have to rebuild these manually, unless there is already a
relevant repo with these rebuilt on Rocky/Alma/etc., which I am not
aware of.
If you cannot use RHEL, Centos Stream 8 is the only option. This is the
version we test on in ovirt system tests.
I don't know about any issue with Centos Stream 8 for engine host, but for
hosts you are stuck with qemu-6.0, since qemu-6.1 is broken, and 6.2 is
not available.
Indeed.
Best regards,
--
Didi