Host network is resetted after VM command is issued

Community Support
1

Hello,

I’m using OpenNebula 6.10.0.1 community edition.
Set-up is one front-end node and two hosts node.

I see a strange issue link to the VM lifecycle management.

Whenever I issue a Suspend/Stop/Poweroff command to an instantiated VM the command does not go through and I lose total connection to the host where those VM are instantiated. I need to reboot the host and then I can try to update the VM state.

This happens everytime and even while terminating the instances some times.
I’ve seen that the host reset the static ip address on the nic then I cannot ping/ssh it anymore.

Host OS is Ubuntu 22.04, one nic only.

Is there any kind of log I can observe to get more informations?

Thank you

2

Hello @Manuel_Femia

Uselful logs could be:
/var/log/syslog, /var/log/messages, or /var/log/daemon.log to check status of the VMs.

For OpenNebula logs, you can go and check /var/log/one/oned.log and /var/log/one/sunstone.log

Hope it helps.

Cheers,

3

Hello,

It is a long shot, as there is no much context provided of the network setup. Is the ip address set on a linux bridge managed by the Host OS?
You could try setting :keep_empty_bridge: true in /var/lib/one/remotes/etc/vnm/OpenNebulaNetwork.conf then sync the hosts with sudo -i -u oneadmin onehost sync --force. Then try again…
If that is the case, another workaround is to not define the PHYSDEV attribute on the VNETs where the Host managed bridge is in use…

Best Regards,
Anton Todorov

4

Hello @atodorov_storpool

sorry for the late response. it’s like you’ve told. I have a bridged network.

Will look into the confs you suggested even if now it seems like it’s working better.

Thank you very much.