I don’t understand.
Is oVirt like the host, and then these desktops are real PCs running somewhere else, and then the real PC connects to something like a VM guest on oVirt … and then what?
Using oVirt only. VDI desktops. Have 4+1 HPE blades with 2+ TB RAM.
Have informed RH personnel multiple times my opinion about dropping spice. Have multiple times asked what is the actual reason. Have used multiple hours trying to figure it out. Even found some weird and obsolete Chinese patents. Never ever have had the real answer it’s all about that codec.
//arl
In my opinion, the reason is simple, they simply decided to get rid of oVirt, glusterfs and spice in favor of openshift, because it is more complicated to deploy and therefore more profitable in terms of cash when selling support.
I don’t use oVirt, which is by the way complex to maintain, see how Oracle struggled to provide its own 4.5 update. However I rebuilt ligovirt 0.3.8 that was provided with RHEL 9 BETA, does it work for you ?
Dear all,
I’m new here and I have to admit that I use Alma Linux 9.x, but often follow this forum, because I found very useful information here also for Alma Linux. I hope that I will not cause problems or offend anyone with this. I 'm just looking for solutions to my computer problems.
For quite a long time I use the repo for QEMU KVM with SPICE support from the user ligenix - the repo is enterprise-qemu-spice
Big thank you ligenix for providing these packages!
From the last update of my Alma 9.x system (last week) I’m not able to run any virtual machine any more. I get the following error in my log files:
qemu-kvm: ../qapi/qobject-output-visitor.c:95: void qobject_output_add_obj(QObjectOutputVisitor *, const char *, QObject *): Assertion `name' failed.
2025-01-12 20:37:17.686+0000: shutting down, reason=crashed
System specs:
libvirt version: 10.5.0, package: 7.2.el9_5 ,
qemu version: 9.1.0qemu-kvm-9.1.0-8.el9_spice,
kernel: 5.14.0-503.19.1.el9_5.x86_64
I have already searched the internet and found that there is a bug reported here:
RHEL issue tracker
I would like to know if anyone here that uses the QEMU KVM with SPICE support has also encountered this problem and if he/she already found a solution to run the virtual machines again.
Best regards,
OMMB
The problem is, unless the packages are perfectly aligned with Rocky 9.5, it’s impossible to test. In other words, your qemu-kvm package is not the one from the official Rocky repos?
In upstream land, there is this patch
https://lists.nongnu.org/archive/html/qemu-devel/2024-11/msg05454.html
Dear gerry666uk,
as I wrote I use Alma Linux 9.5 (fully updated) and the problem started last week after an update.
I suppose that you are running Rocky 9.5 with qemu-kvm with SPICE support without any problem?
Best regard,
OMMB
The question is, what updates were applied when the problem occurred? Also:
root@rocky9:~# dnf info qemu-kvm
Last metadata expiration check: 2:56:51 ago on Thu 16 Jan 2025 06:24:52 AM CET.
Available Packages
Name : qemu-kvm
Epoch : 17
Version : 9.0.0
Release : 10.el9_5
Architecture : x86_64
Size : 71 k
Source : qemu-kvm-9.0.0-10.el9_5.src.rpm
as you can see qemu-kvm version in Rocky 9.5 differs to Alma (you have 9.1.0). Alma decided to be binary compatible, so therefore it seems their packages do not match Rocky nor RHEL anymore.
But as spice is dropped in RHEL9, it’s also dropped in Rocky9 as well. You would probably be better asking this question on the Alma forums because since the version of their packages differs to Rocky I’m not sure how people who use Rocky can help you. It’s almost like asking for help on a Debian forum for Ubuntu when the package versions differ between the two.
Dear iwalker,
I understand your point. Rocky is not Alma.
I write to this forum, because only here I have found some users that (at least in the past) used an external repository called enterprise-qemu-spice provided by this forum user ligenix which is hosted on Fedora COPR. The packages there are of different versions as from Alma official repositories (and also Rocky).
I just wanted to know from this thread forum users (which perhaps are still using the packages from enterprise-qemu-spice) if they have encountered the same problems like me running virtual machines with qemu-kvm with SPICE support on Rocky because Rocky and Alma are very similar (in my personal opinion almost the same, please do not be offended by my opinion).
I will try to contact the mantainer of enterprise-qemu-spice through Fedora COPR system, although I’m not very familiar with Fedora COPR system.
Thank you very much, for taking your time and replying to my posts!
Best regards,
OMMB
1 Like
Dear ommd, gerry666uk and iwalker,
Thanks for the report, the patch link and theses useful comments. The problem occurred after the latest patches Red Hat added in qemu-kvm for CentOS Stream 9.
I just recompiled the qemu-kvm RPM with the patch provided in the link and successfully updated my portable server.
2 Likes
Dear ligenix,
I can confirm that the last updated version of qemu-kvm works again:
Installed Packages
Name : qemu-kvm
Epoch : 18
Version : 9.1.0
Release : 9.el9_spice.1
Architecture : x86_64
Size : 0.0
Source : qemu-kvm-9.1.0-9.el9_spice.1.src.rpm
Repository : @System
From repo : copr:copr.fedorainfracloud.org:ligenix:enterprise-qemu-spice
Thank you very much for a quick problem resolution and providing us the SPICE support for qemu-kvm EL9!
Best regard,
OMMB
1 Like
Dear ommb,
I use Rocky Linux daily on my laptop and servers, and I am grateful to this forum for allowing me to provide some support for the COPR qemu-kvm-spice repository.
I have unfortunately been very busy, both professionally and personally for the past few weeks, but I am trying my best to overcome this lack of availability.
Thanks for this positive feedback.
Best regards,
ligenix
2 Likes
Hi,
glad to find this, I have a Rocky 8 server with a RHEL 5 VM and a CentOS Stream 8 workstation with three Windows VMs. All VMs have SPICE/QXL graphic cards and that stopped me from upgrading my ws to Rocky 9,
However, what stops me most is that all VMs have this chipset
<type arch='x86_64' machine='pc-i440fx-rhel7.0.0'>hvm</type>
and RH also dropped support for them,
Do these packages ligenix built also support the same chipset RHEL8 supported ?
Changing chipset will probably blow all licenses in Windows, XP, 7 and 10, and I need that to stay working to serve several LTS projects,
Thx
Hi scarpile,
I have succefully migrate a Windows 2012r2 for CentOS 7 to Rocky 9.
Please note that the chipset needs to be updated as follows:
<type arch='x86_64' machine='pc-i440fx-rhel7.6.0'>hvm</type>
which doesn’t seem like a major step.
Feel free to test either.