Re: libvirt backport to Bookworm

Guido Günther <[email protected]>
Newsgroups gmane.linux.debian.backports.general
Message-ID <[email protected]>
Hi Thomas,

On Wed, Jun 19, 2024 at 05:48:33PM +0200, Thomas Goirand wrote:
> Hi there!
> 
> We're about to put in production a 2nd region of our public cloud, which is
> OpenStack based, and using libvirt / kvm.
> 
> The hardware we have is rather recent: each compute is equiped with 2x EPYC
> 112 core CPU. That is the EPYC-Genoa CPU model in Qemu.
> 
> Therefore, we've installed the latest qemu from bookworm-backports. However,
> libvirt is still too old, and doesn't know about this CPU. Therefore, nova
> thinks it's model is EPYC-Milan, and not EPYC-Genoa.

Did you check if updating the cpu map in /usr/share/libvirt/cpu_map/ is sufficient?

> I've done a backport myself of python-libvirt and libvirt, and it's painless
> to do: just a simple rebuild without changing the package.

Glad that worked so well.

> 
> Therefore, I'd like to ask: does the libvirt team (guido?) would like to
> upload a backported version of libvirt and python-libvirt to Bookworm? If
> yes, that'd be great. If no, would you mind if I do the upload? Will you
> help maintaining it?

I wouldn't have much bandwidth to help here. Let's see if Andrea has any
objections as he's doing most of the work in sid/trixie atm.

Cheers,
 -- Guido

> 
> Note that it's working perfectly well for me: I was able to spawn VMs, and
> live-migrate them between compute hosts running the latest OpenStack Caracal
> version.
> 
> Cheers,
> 
> Thomas Goirand (zigo)
>
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.