Hello,

I did something like that via "virsh edit HostedEngine".

But how is the change written back to the shared storage ("hosted_storage" so it stays permanent  for HA Engine ?

I was able to boot up HostedEngine manually via virsh start after removing the required flag from XML (I first added a user to sasldb in /etc/libvirt/passwd.db to be able to log into libvirt).

Bye

Am 28.11.2019 um 05:51 schrieb Strahil:

Hi Ralf,
When the deployment fail - you can dump the xml from virsh , edit it, undefine the current HostedEngine and define your modified HostedEngine 's  xml.

Once you do that, you can try to start the
VM.

Good luck.

Best Regards,
Strahil Nikolov

On Nov 27, 2019 18:28, Ralf Schenk <rs@databay.de> wrote:

Hello,

This week I tried to deploy Hosted Engine on Ovirt-Node-NG 4.3.7 based Host.

At the time the locally deployed Engine ist copied to hosted-storage (in my case NFS) and deployment tries to start the Engine (via ovirt-ha-agent) this fails.

QUEMU Log (/var/log/libvirt/qemu/HostedEngine.log) only shows "2019-11-27 16:17:16.833+0000: shutting down, reason=failed".

Researching the cause is: The built Libvirt VM XML includes the feature "virt-ssbd" as requirement, which is simly not there.

From VM XML:

  <cpu mode='custom' match='exact' check='partial'>
    <model fallback='allow'>EPYC</model>
    <topology sockets='16' cores='4' threads='1'/>
    <feature policy='require' name='ibpb'/>
    <feature policy='require' name='virt-ssbd'/>

from cat /proc/cpuinfo:

processor       : 47
vendor_id       : AuthenticAMD
cpu family      : 23
model           : 49
model name      : AMD EPYC 7402P 24-Core Processor
stepping        : 0
microcode       : 0x830101c
cpu MHz         : 2800.000
cache size      : 512 KB
physical id     : 0
siblings        : 48
core id         : 30
cpu cores       : 24
api