Il giorno ven 4 feb 2022 alle ore 09:52 Thomas Hoberg <thomas(a)hoberg.net>
ha scritto:
> On Tue, Feb 1, 2022 at 7:55 PM Richard W.M. Jones
<rjones(a)redhat.com>
wrote:
>
> Would you like to file a doc bug about this?
>
> oVirt on RHEL is not such a common combination..
Well IBM seems bent on changing that (see the developer license post below)
>
> In CI we only test on Centos Stream (8, hopefully soon also 9, we'll
see),
> and RHV on RHEL.
Which is exactly why BetaStreamOS (upstream to RHEL) based oVirt is
breaking the value proposition of TrueCentOS (downstream of RHEL) based
oVirt as a solution "for the entire enterprise".
AFAIK there is no "developer license" of RHV so if you want to run a lab
environment where the experimentation is done at app or science level (not
OS), this new beta(OS)-on-beta(VM orchestration) stack just becomes too
fragile to use, not that oVirt HCI has ever proven better than hardware
level fault resilience to me, which I believe was its design goal.
> We'll be happy to get success/failure (with details) reports of attempts
> to install/setup both engine and hosts on other OSes.
>
I have been working on just that, going through Alma, Rocky and Oracle
(RHEL, Liberty and VzLinux might be added) using a nested VMs on VMware for
two scenarios:
1. Installing oVirt on a new hardware node (a CentOS8 VM switched to each
new EL8 brand)
2. Switching a pre-existing oVirt HCI node to a new EL8 brand
In all cases the virtual hosts wind up running a single node HCI gluster,
to keep things reasonable simple.
All of these worked fine,with some tiny glitches, but that was before the
gluster8 repos got archived. Currently I am awaiting that being fixed to
continue testing and hopefully posting a somewhat bigger report here.
But there is another new issue that bothers me quite a bit: The oVirt
management appliance:
I managed to still get one of the latest CentOS8 based appliances during
my tests and that could evidently be switched to one of the other EL8
variants, once the repo issues are sorted out.
But newer versions of the appliance are built on the UpStreamBeta, which
cannot (easily?) be switched to a downstream EL8, while the RHEL appliance
most likely can't be used legally.
Using a Beta engine without the RHEL benefit of vulnerability management
is a no-go in PCI-DSS or similar compliance heavy environments, so for
oVirt to be a "solution for the entire enterprise" there needs to be an EL8
based engine, too.
But AFAIK there is not even a publicly available build script for the
appliance so someone outside RH could build a matching bot. And forced
downgrades of the engine to one of the EL8 variants might well break the
engine...
All the build scripts for oVirt Appliance are publicly available:
https://github.com/oVirt/ovirt-appliance
If you want to rebuild an appliance with Rocky instead of CentOS Stream you
can just download this source then:
- adjust repos in
https://github.com/oVirt/ovirt-appliance/blob/master/engine-appliance/dat...
to math Rocky's
- adjust makefile pointing to Rocky's boot iso in
https://github.com/oVirt/ovirt-appliance/blob/master/engine-appliance/Mak...
- run make rpm.
> Next step is probably for someone to try and build an oVirt node image
> based on a different OS...
I would love trying (if I can find the time...), but I'm not sure where I
could find build scripts...
For oVirt Node it's almost the same from
https://github.com/oVirt/ovirt-node-ng-image
adjusting repos in
https://github.com/oVirt/ovirt-node-ng-image/blob/master/data/distro-defs...
and boot.iso in
https://github.com/oVirt/ovirt-node-ng-image/blob/master/configure.ac#L46
or just passing it to configure command line.
We have no capacity to build and test for all the variants but there's no
secret on how to build node and appliance.
I'm pretty sure I was discussing this with someone either on Alma or Rocky
chat about a year ago and kindly asked to reach out to
$preferred_distribution
to build oVirt Node and ship it in their mirrors too.
One of the reasons I could never use these node images was that the
2.5Gbit USB Realtek driver is broken in EL7+EL8, which I used on my Atom
based 3 node HCI. Building node images would allow adding drivers and
putting in nested virt support...
>
>
> Perhaps I am missing something - is this oVirt-specific, or
> relevant?
>
> Thanks and best regards,
_______________________________________________
Users mailing list -- users(a)ovirt.org
To unsubscribe send an email to users-leave(a)ovirt.org
Privacy Statement:
https://www.ovirt.org/privacy-policy.html
oVirt Code of Conduct:
https://www.ovirt.org/community/about/community-guidelines/
List Archives:
https://lists.ovirt.org/archives/list/users@ovirt.org/message/ELW5CNTXJNA...
--
Sandro Bonazzola
MANAGER, SOFTWARE ENGINEERING, EMEA R&D RHV
Red Hat EMEA <
https://www.redhat.com/>
sbonazzo(a)redhat.com
<
https://www.redhat.com/>
*Red Hat respects your work life balance. Therefore there is no need to
answer this email out of your office hours.*