@olivierlambert Thank you for your reply. (You're very active and responsive in forums. I noted your replies to Tom_LTS in his videos last year!)

I actually don't think it is a huge issue right now. Xen is running, and I am able to get in and set things up via ssh into dom0 at the moment. Some of the problems I run into are just due to lack of / outdated documentation on Xen itself. I'm quite technically capable (despite being frazzled trying to get such a complex system setup going), but the stumbling blocks are the documentation on the internet, and a lack of a good venue to ask clarifying questions. It's been a while since I've used IRC and mailing lists, and that seems to be the only/best source for Xen community technical questions.

I've looked into Qubes, and it seems to be compelling, but I want to try to build this path from the ground up first. If it proves to be too much hassle or is unstable and I can't get any further with it, I can always drop back and go the Qubes route.

Are there any other, more modern community venues to get Xen-on-Desktop technical questions considered? I'll go as far as I can with the available online documentation, and google the exceptions as I run into them (like this one), but it would be nice to find a place where I could go to ask the questions my google-fu doesn't suffice to locate answers.

Anyway, thanks again for the information/assistance! I do have plans to use XCG-ng on some of my servers once I get things all set up in the office for them.

Greetings,

I realize that this forum is primarily for XCPng, and I apologize in advance if general Xen/Desktop questions are out of place.

I've been planning a move to Xen for my desktop for a while now, after researching various other alternatives, like KVM, ProxMox, Unraid, VMware, etc. I wanted to go with a Type 1 "bare metal" hypervisor and build up a set of guest vms I can spin up as needed, but primarily have a Windows 10 guest as my daily driver, with a Linux desktop guest for some of my Linux dev work. I do plan on doing passthrough for a lot of the hardware into Windows, unless and until I can get vGPUs (GPU partitioning) capability going.

I started getting things set up a few months back with a new SSD and cleaning up space on an NVMe SSD to act as my core image store and Xen boot volume. I am using CentOS 7 as my dom0 and followed the available guides (Xen4CentOS/Xen4QuickStart) to get both the OS and Xen installed and set up. However, at some point, I ran into a problem I could not get past, so had to shelve the migration until now. My older video card finally gave up the ghost, and I decided to take advantage of the interruption in productivity to just bite the bullet and work through to get this system up on Xen.

The problem I have run into is this: when I try to boot the Xen-enabled kernel from GRUB, I get the message "WARNING: no console will be available to OS" message from the xen layer, and the last message that appears before the screen blanks is the initramfs load. From the logs, I can see that Xen/CentOS does finish booting and seems to be running fine, but it doesn't give me a console.

I have tried googling everything I can think of about GRUB2, xen command line params, and CentOS kernel command line parameters regarding that message, and have fiddled with it for hours on end at this point, but nothing I do seems to change this or provide any other feedback to point me in a direction to go in. Most of the results point to adding things like vt.handoff which is an ubuntu-kernel-specific parameter, and does nothing on CentOS.

My system is an Intel Core i7-5960X (8C16T) with 128GB of memory, and plenty of SSD and rotating rust space. I have configured the system to use UEFI to boot, and the Xen partition is GPT (I migrated from Legacy boot/MBR) back when I started this months ago). I fixed a few issues with GRUB video modes (setting to gfxmode with screen resolution). The new video card is a Radeon RX6950XT 16GB.

If possible, if I can get past this current issue, I'd like to get some recommendations about what is presently possible with vGPU/GPU-P support (I've seen some recent posts on Phoronix about it), and what I should be able to fall back on in terms of a working GPU passthrough config in the meantime. Ultimately, I'd like to share the GPU, but I am good with using passthrough for Windows and virtual desktop (ie, Parsec/Remmina/etc) for Linux desktop (and test VMs) for the time being.

I realize that there may be some issues with dom0 contention with a single GPU passthrough to a Windows domU, and I'll burn that bridge when I get to it once I get this elephant problem out of my way.

Thank you in advance for any assistance, and sorry to bother if I am barking up the wrong tree.