On Sun, Dec 12, 2021 at 2:58 PM Yedidyah Bar David <didi@redhat.com> wrote:
Hi,
On Sat, Dec 11, 2021 at 5:38 PM Paul-Erik Törrönen <poltsi@poltsi.fi> wrote:
I have a server running the engine on CentOS Stream.
2 hosts running likewise CentOS 8 Stream (C8S) and 2 hosts running Rocky Linux 8.5 (RL8.5).
I had several VMs running on the 8S without any issue. After the oVirt upgrade, which partially fails* on the C8S-machines, the VMs on the C8S-hosts appear to be running, according to the oVirt GUI, but I can neither ping, ssh or even open a virt-viewer console to the VM. The console shows just a black screen.
The only thing shown in the GUI is a note that the guest agent should be updated on the VM, so there appears to be some kind of communication between the engine/host and the VM.
If I create a new VM on the C8S, I get the same result: No access to it, console is black despite the fact that oVirt GUI shows the VM as running and green.
Perhaps it's due to qemu-6.1.0? Please see the list archives - try to downgrade to 6.0.0.
If I create a VM on one of the RL8.5 hosts, the console does work, I can eg. complete the GUI-installation (from CD) there. But after the boot, despite having configured the network correctly in the oVirt-GUI and in the installation process, the VM (using the console) is unable to raise the network.
No error is displayed in the oVirt GUI neither for the cluster, DC, host nor VM.
No apparent log-error neither in the engine-server nor the hosts.
Stopped the firewalld on both in the engine-server as well as the hosts, but this did not change anything.
It seems that the 4.4.9-upgrade somehow broke the oVirt logical networking, completely on C8S-hosts.
I think something like this was also mentioned on the list recently, do not remember details.
* The C8S hosts fail on update WRT the following packages:
Please try 'dnf update ovirt-release44' and then retry.
See also: https://bugzilla.redhat.com/show_bug.cgi?id=2023343 Best regards, -- Didi