The issue seams here: please ensure that you can correctly connect
your storage server.
Can you please attach vdsm logs?
Yes actually I figured out it was a DNS problem : as mentioned in the messages from the
log I provided, it wasn't able to reach the NFS where the engine was (as it uses FQDN
not IP with NFS it seems, I will fix that for not depending on DNS).
This is actually my setup : only Em1 is plugged, it has ovirtmgmt + one other logical VLAN
network. This VLAN network as in DHCP and never had an IP, everything was working fine.
Since I added an IP address to that interface, the manager crashed. Actually it is trying
to use that VLAN interface as the default route, I have no idea why, and cause DNS issue
(one of the DNS was on another network, the the second was on the game network...it should
actually have worked anyway...).
The only way I found to resolve this was ifdown of that interface, and route add default
gw <gateway-IP> ovirtmgmt
After that, I had errors like "unknown stale data" and "failed to
reinitilize lockspace" ; removing the lockfile with hosted-engine command, and
removing manually __DIRECT_IO__ file on the engine storage didn't fix it.
I actually found out what was happening : ovirt-ha-agent had errors in his status (with
systemctl), ovirt-ha-broker had errors related to ha-agent and vdsdm had errors related to
those 2 previous services.
I resolved my issue by restarting the service in the good order :
# systemctl restart ovirt-ha-agent.service
# systemctl restart ovirt-ha-broker.service
# systemctl restart vdsmd
Anyway thanks for your answer, I hope this topic will help people with similar issues