I just created a new cluster using Intel NUCs (i5 processors with 16GB). I was using this earlier to deploy VMs using OpenNebula over on Debian 12. Currently deployed on Ubuntu 24.04 LTS.
Datastore is local and on the same server/host as the frontend.
I downloaded Fedora 42 image from the marketplace and tried to deploy it. First it got scheduled on nuc3. As the VM deployment did not complete and was in “BOOT” state for too long, I tried to see if I can ssh to nuc3 but it was not reachable at all. I had to go physically power cycle it to come back up. Then same thing happened to nuc2. I thought i’ll try something smaller, i deployed Alpine that got scheduled on the same host on which the front end runs. Now the frontend nuc1, nuc2 and nuc3 are all unreachable.
Can someone please point me in the right direction to fix this issue.
Thanks.
Ajay
Versions of the related components and OS (frontend, hypervisors, VMs):
Host OS Ubuntu 24.04
OpenNebula 6.10 (latest)
Steps to reproduce:
Just try to deploy a VM
Current results:
Hosts on which the VMs were deployed became unresponsive and unreachable. Power was still on but had to be powercycled. Expected results:
It should easily be able to run a few VMs with no other workloads.
Have you created those VMs with a network interface? Have you tried to create them without network? Do they work without a network connection?
If you cannot access to the host via SSH after you create it and any VM stays on BOOT state, the problem may have to do with the bridge creation over the same interface that the host is using for SSH.
The easiest way to ensure the network configuration on the hosts is using netplan. Creating the bridges with NetworkManager as well may work
I suspected the bridges as well. I did not realize I have to create the bridge myself. Thought the GUI will do it for me when i created the virtual network.
Can you please point me to any documentation of examples of creating this. I know how to create bridges on Linux. What I don’t know is how OpenNebula GUI works with it.
Also these VMs are stuck in Boot state. Only thing I’m allowed to do with them is “Recover” which does a “retry” and then i’m back to host being unreachable. How can I get out of this and clean up so I can start again?
About bridges, OpenNebula needs one for each Virtual Network. If you don’t provide one, OpenNebula will create its own, but it will quite simple and only allow local host networking.
The best solution for a small setup without a dedicated NIC for the vnets probably is VXLAN with BGP-EVPN (even if it looks overkill). Please, check the Using VXLAN with BGP EVPN section on the docs. Also remember that VXLAN will add info on the VM packets, so set the MTU of the VMS as the MTU of your hosts -50bytes (1450 bytes should be OK)