@stormi Nice to do something useful and give a little back, other than buying a XCP-ng cushion for my office!

@r1 Is this still in development; would like to get it working with 8.1?

Manually mounting the CephFS works, but the patch fails, probably not surprisingly; or is a GlusterFS approach a better option?

With Ceph Octopus and the cephadm approach setting Ceph up is really easy, so maybe this will become a more popular feature....?

Windows Home is 1 socket, Pro is 2 and Workstation / Enterprise is 4 apparently......

Windows Server is many more 64 apparently, currently I have 12 dual core sockets on my VM.

@wsciaroni just double checked, I see that too very early on and not for long, but then it boots.

Just downloaded this iso - http://releases.ubuntu.com/18.04.4/ubuntu-18.04.4-live-server-amd64.iso

Works on my E5-2697v3 and E5-2620 v4 both on a Supermicro X10DRi-T, (only one CPU in exch).

So after parking CoreOS for nearly a year, and after noticing some edits and commits had been made I decided to give it another go; this time with success…. Yipppeee

• XCP-ng 8.0.0 with latest patches
• CoreOS version is Container Linux 2303.3.0
• XO from sources - xo-server 5.54.0

There are a few buts and pointers though!!

Deploying the VM did not work from XO (from sources, just a homelabber here); the config disc was not created.

So reverting to XCP-ng Centre…….

Following https://github.com/xcp-ng/xcp/wiki/Docker-in-XCP-ng for general guidance – BUT I was unable to log in with my SSH key.

This is fixed by enclosing the key in “ “ thus, my cloud-config was

#cloud-config

hostname: %VMNAMETOHOSTNAME%
ssh_authorized_keys:
  - “ssh-rsa AAAAB3NzaC……………………..syQ4cJYnG70BqQJZVqo5fGcIQP9j5UY1 user@host"
  # - ssh-rsa <Your public key>
  # The following entry will automatically be replaced with a public key
  # generated by container management plugin. The key-entry must exist,
  # in order to enable container management for this VM.
  - ssh-rsa %CONTAINERRSAPUB%
coreos:
  units:
    - name: etcd-member.service
      command: start

    # Hypervisor Linux Guest Agent
    - name: xe-linux-distribution.service
      command: start
      content: |
        [Unit]
        Description=Hypervisor Linux Guest Agent
        After=docker.service

        [Service]
        ExecStartPre=/media/configdrive/agent/xe-linux-distribution /var/cache/xe-linux-distribution
        ExecStart=/media/configdrive/agent/xe-daemon
  etcd:
    name: %VMNAMETOHOSTNAME%
    # generate a new token for each unique cluster from https://discovery.etcd.io/new?size=3
    # specify the initial cluster size using ?size=X
    # discovery: "https://discovery.etcd.io/<token>"
write_files:
  # Enable ARP notifications for smooth network recovery after migrations
  - path: /etc/sysctl.d/10-enable-arp-notify.conf
    permissions: 0644
    owner: root
    content: |
      net.ipv4.conf.all.arp_notify = 1


# Template loaded from /usr/lib/python2.7/site-packages/xscontainer/data/cloud-config.template

Following the rest of the guide everything appeared to go well.

Launching a container and then the next issue, it doesnt show in XO or XCP-ng Centre. Following this post https://xcp-ng.org/forum/topic/1679/how-to-fix-coreos-docker-container-tree-view-in-xcp-ng-center/2 and vola, all working.

Now I have a working mater a fast clone avoids any of these issues for additioanl hosts.

Creating a container without –restart always does not allow me to start / restart / stop it from XO, not the end of the world, looks like this has been found https://xcp-ng.org/forum/topic/2148/can-t-start-docker-container-from-xoa

Happy to try and help resolve any of these 'issues', test and provide logs etc. etc.

@geek-baba Fair enough, just another option. I don't like block storage, always favoured NFS over iSCSI, i think the speed trade off is worth the transparency.

@geek-baba or a CephFS option!

@imtrobin No, AMD. Currently a WX4100 (as its LP and fits into a 2U server) but have also had a RX580 working before.

Windows Home is 1 socket, Pro is 2 and Workstation / Enterprise is 4 apparently......

Windows Server is many more 64 apparently, currently I have 12 dual core sockets on my VM.

@linhnb said in XCP-ng GPU Passthrough - No GPU found:

You can use the GUI. I checked, the code changes as if you do make it in CLI. please check the newest version 8.1

Thanks for the tip!! Working in 8.1 in XCP-ng Center and also XO

Also, dont think there is a need to 'hide' the GPU from Dom0 now??

edit: Did a fresh install, no need to 'xen-pciback.hide' when assigning via XO or XCP-ng center. This was with AMD as they play nicely, not sure about NVIDIA.

If only they made a LP vGPU card.... Looks like a WX4100 will have to do.

@linhnb don't think you can use the GUI for GPUs, only vGPUs.

RX580 does work, just follow the PCI passthrough guide.

@anthonyh Google code 43 NVIDIA...... Basic translation is we want to stop you doing this with a consumer card in a VM. There are solutions I believe.

Just to report back..... So far so good. I've moved over a few VDIs and not had any problems.

I've rebooted hosts and Ceph nodes and all is good.

NFS is also all good now.

Hope this gets merges soon so I don't have to worry about updates

On a side note, I've also set up two pools, one of SSDs and one of HDDs using File Layouts to assign different directories (VM SRs) to different pools.

Just checked uefi too, that also works.