Thanks for the reply.

I had already checked what you suggested. I didn't realize that the engine was used only for management, I had assumed that if it crashes it would take down all the VMs.

ovirt-ha-broker seems to have crashed. In the borker.log I see:

RuntimeError: Couldn't  connect to VDSM within 60 seconds
Thread-3::ERROR::2022-06-16 08:35:02,857::submonitor_base::119::ovirt_hosted_engine_ha.broker.submonitor_base.SubmonitorBase::(_worker) 
Error executing submonitor mem-free, args {'address': '0', 'use_ssl': 'true'}

hosted-engine --vm-status tells me that the agent has stopped.

If I do "hosted-engine --vm-start" what will happen to the running VMs ? Since it's only used for management, if it detects that they are already running, will it ignore them and go on it's merry way ? I'm trying to avoid shutting down all the servers if I can help it.

On 6/15/22 11:07 PM, Sketch wrote:

On Wed, 15 Jun 2022, Valerio Luccio wrote:

I have an ovirt 4.4 installation whit self-hosted engine where the agent
seems to have died. The VMs are still running, so I assume that the engine
itself is still running (is this a wrong assumption ?). Can I restart the
agent without affecting the running VMs, that is how will restarting the
agent affect the running VMs ? If I can restart the agent, what's the
correct way of doing it ?

If the engine is down, the VMs will continue to run.  You just won't be able to start/migrate/configure/etc them.

If the engine VM is still running, you may want to SSH into it and look at the state of the system to see if you can see what went wrong.

systemctl status 'ovirt*' may tell you if just a single service is down (such as ovirt-engine).  You might also check the logs in /var/log/ovirt-engine

If the VM is down or inaccessible, SSH into one of the hosts capable of running the engine and run the following to check VM status:

hosted-engine --vm-status

This should tell you if and whre it's running.  If it isn't dead, You can stop it with:

hosted-engine --vm-shutdown

Check the status and wait until it's actually down, then you can start it up again:

hosted-engine --vm-start