@olivierlambert perfect. Thanks. Am currently doing exactly that.
Had to fiddle a bit with the hetzner vSwtitch Firewall, if I find the time I will post some instructions that could maybe help others as well.

@Forza thanks. But as @normanghenderson a live migrate from 8.2 -> 8.3 didn't work for me either.
I tried that approach first, made the 8.3 to a new pool but it delivered some XCP API Errors ... which thinking of it now might have been a firewall issue but I manually checked access between the systems and everything seemed fine.
I now have everything running on 8.2. Have some non important VM on the new server that is 8.2 and will upgrade that one to 8.3 then see if I can migrate other VM away from the 8.2 systems.
If that works, I'm all good. If not ... I will have to start over from scratch.

But thanks everyone for the quick replies.

Hi xcp-ng community,
I've have a xcp-ng setup consisting out of a pool of two servers that are both running XCP-ng 8.2.1, yes yes, I know it's outdated and I should have upgraded. Don't judge please, it's been a rough year.
Now I've finally made time to setup a third server so I can migrate and upgrade the others. This is where my problems started, I can't add the new xcp-ng running 8.3.0 to the existing pool.
I'm getting an error "message": "The hosts are not compatible",
Which I assume is based on the fact that they are running different xcp-ng versions.

I was thinking of two paths out of this mess.

1. Reinstall the new server with 8.2.1, adding it to the pool then migrate VMs and start upgrading.
2. Upgrade the existing servers then add the new one to the pool.

The 2nd option sounds like the more reasonable idea but it just feels scary to install an iso over an existing system.

Was just hoping to be able to move the VM off to the new server and start a rolling upgrade but since I can't add it to the pool it seems that I have to take the risk.

Any suggestions how I could do this better? Or what to pay attention to?

Thanks

@olivierlambert weird, I tried the restart, because I'm doing that a lot now since I have some backup issues.
Need to move my old (very old) datacenter to the new XCP-NG Server and after that I can get to upgrading XO.

I might be running into this issue, not sure however, will upgrade my xen Orchestra soon to confirm
https://github.com/vatesfr/xen-orchestra/issues/5896

but for now I would just like to know how I can cancel these tasks. I have two of them running, cancel and Disabled buttons are not active.

Here's what I did. I added a new Physical machine, so far I have only moved some VM to that machine from the first one.
Now for the first time I created a new Machine with the Debian 11 cloud template, and the tasks that are stuck [XO] Importing content into VDI XO CloudConfigDrive (on x1.example.com) 0% show x1 which is my first physical machine. I have no idea why it would be running there.

RobertH1993 created this issue in vatesfr/xen-orchestra

closed [BUG] Adding new cloud-init VM causes hanging VDI imports. #5896

@olivierlambert Thanks, just stumbled over this a few minutes ago and tried it, but doesn't work. According to https://www.digitalocean.com/community/tutorials/an-introduction-to-cloud-config-scripting#write-files-to-the-disk

This currently only works for RHEL-based distributions.

But this led me to write_files and this works:

write_files:
  - path: /etc/resolv.conf
    content: |
      domain exmple.com
      search example.com
      nameserver 192.168.20.1

this is the full configuration I use:

#cloud-config
hostname: {name}
timezone: Europe/Vienna
users:
  - default
  - name: johndow
    passwd: ****
    sudo: ALL=(ALL) NOPASSWD:ALL
    groups: users, admin, sudo
    shell: /bin/bash
    lock_passwd: true
    ssh_authorized_keys:
      - ssh-ed25519 AAAA****
write_files:
  - path: /etc/resolv.conf
    content: |
      domain example.com
      search example.com
      nameserver 192.168.20.1
package_update: true
packages:
  - sudo
  - htop
  - vim
  - python
  - tmux
package_upgrade: true
runcmd:
  - sudo reboot

the Network configuration looks like this:

  version: 2
  ethernets:
    eth0:
      match:
        name: eth0
      addresses:
        - 192.168.20.6/28
      gateway4: 192.168.20.1
    eth1:
      match:
        name: eth1
      addresses:
        - 10.10.10.6/24

but now it seems I forgot the MTU but this is a different story, thanks for the hint in the right direction.

@olivierlambert sure ... how?
I couldn't find out how to do that. Looking into that option right now but if you have some hints I would appreciate it.

@olivierlambert it's not that simple I guess. Because I can't install anything without having DNS during the cloud-init
Everything works fine if I use DHCP but this would result in eth0 being my internal network and default GW being via the internal router and eth1 having my public IP.
I want it the other way round

The second solution would be to use DHCP to get everything installed then nuke the whole cloud-init stuff and configure network manually. Bit of a pain I guess.
How are other people dealing with this whole clod-init on debian?

I've read this post and also been trying my own research but I just can't get it to work.

https://xcp-ng.org/forum/topic/6777/cloud-init-network-config-not-applying-to-debian-11-6?_=1692953942818

This is my network cloud init configuration

network:
  version: 1
  config:
    - type: physical
      name: eth0
      subnets:
         - type: static
           address: 192.0.2.6/28
           gateway: 192.0.2.1
    - type: nameserver
      interface: eth0
      address:
        - 9.9.9.9
        - 1.1.1.1
      search:
        - example.com
    - type: physical
      name: eth1
      subnets:
         - type: static
           address: 10.10.10.6/24

The result is this file data in /etc/network/interfaces.d/50-cloud-init

auto lo
iface lo inet loopback
    dns-nameservers 9.9.9.9 1.1.1.1
    dns-search example.com

auto eth0
iface eth0 inet static
    address 192.0.2.6/28
    gateway 192.0.2.1

auto eth1
iface eth1 inet static
    address 10.10.10.6/24

how do I get the network in such a way that DNS also works? If I add my nameservers to /etc/resolv.conf everything works.
I'm also a bit confused as why I have eth0 as dhcp on in /etc/network/interfaces but nothing regarding eth1

This cloud-init stuff is driving me bananas!

@florent perfect! Thank you.
I want a clean start since the Backup Server had some issues and ran out of space so I just want to start fresh.

I want to start fresh with my backups.
I have removed all backup jobs from XOA, all backups through the Health tab. Only Backup left is the Metadata.
However on the Backup NFS Server I still have 350GB worth of files under the xo-vm-backups directory.
Can I safely delete them? Or what's the best way to safely remove all backups and start from Scratch?