Swtich interface (vlan) on xcp-ng vm?

When running an xcp-ng 8.2 server (single host) at home I virtualized xo on a VM. I reworked my network and added new VLANs. I had a catch-22 where I needed to update the VM interface with the new VLAN. XO old VLAN was 110 new one is 28 both networks existed for a period of time. They had the same /24 block of addresses as well. I verified the 28 network was trunking to the xcp-ng host by creating a new VM and assigning its interface to VLAN 28. It worked just fine. I moved all VMs to the new VLANs except XO.

I then moved the XO and was crossing my fingers that it would update itself before dying. I was wrong. It didn’t and was still on 110. Verified via the CLI.

I figured I’d give it a go trying to do it via the CLI. I added a second interface in the 28 vlan and at this time the old 110 network no longer had anything on it besides the xo host. I was thinking maybe xo only dealt with device 0 so I deleted both vifs and then created a new vif on VLAN 28 with the original mac address and the correct network xe vif-create vm-uuid-VM_ID network-uuid=NET_ID mac=MAC_ADDRESS device=0 When rebooting the XO VM I don’t see it’s traffic on the network or it’s mac address in the switch. Looking at the network param list I’ve verified the vif uuid is in the VIF list for the 28 network.

Did I just hose my XO VM? Is there something I’m missing about moving a VM from one VLAN to another?

I know this isn’t the recommended method, however can you use xen center to connect to you xcp-ng interface?

I spun up a new xo in VirtualBox and connected it to the xcp-ng host. Turns out the XO VM system “lost” its eth interfaces. Probably with my delete and adding the other one didn’t show up for some reason. Trying to avoid spinning up a temp XO VM on my local desktop and being “smrt” with xe commands bit me in the end. Next time I’ll just get an XO VM to reconnect to the host and work from there.

Thanks for the suggestion.