I’ve been putting this off for a bit but I did get new servers and thought I would move from the Rock-pi 4SE that has done most of the work at home the last few years. However Iäve put this off for a bit. Today I needed to install a new Access Point for the WiFi and for whatever reason the AP did not show up in the management interface. I first thought the AP modle might have something to do with it. But then I found that the version of the management software I ran was old. Rebooted the container and updated it, no luck. Found out that linuxserver has stopped supporting that container and moved on. The new version requires a monoDB-server. Tried to install that but it turns out mongo no longer supports the architecture of the CPU. Well oh, might as well move it then I thought to myself. Installed a new VM, ziped and transfered data and config, installed docker and try to run it. No such luck. What in the world now? Looking at the logs the (virtual) CPU does not have AVX-support. Then found this and created a new CPU type that supports it. For anyone wondering this is the important part:
New models can be defined here cluster wide:
/etc/pve/virtual-guest/cpu-models.conf
Add this definition to the /etc/pve/virtual-guest/cpu-models.conf file:
cpu-model: x86-64-v2-AES-AVX
flags +avx;+avx2;+xsave;+aes;+popcnt;+ssse3;+sse4_1;+sse4_2
phys-bits host
hidden 0
hv-vendor-id proxmox
reported-model kvm64
It will then be immediately available in the GUI as a Custom (QEMU) CPU
Finally we have a Mono-server that works and that the controller can connect to, and thus start. Then came the “trying to adopt failed”. I guessed this had to do with my controller having a new IP as I needed this to transfer the files. I logged into the PI and shut off the network interface, it runs on Wifi as well so that’s fine for now and change IP on the server. Finally, adoption starts working. During this whole process I’ve been a bit scared as I’ve messed up the controller before and had issues getting the network back up and running but this time I figured it out before the network leasese ran out. Now I finally have the controller installed in a VM that sists in a proxmox cluster. Thus I will be able to update the underlying OS without risking anything as I can migrate the VM over first. Pretty sweet! Next up is probably home assistant which will be totally reinstalled as none of the lables any longer makes any sense as we’ve moved and they were tied to rooms in the apartment.