[FEAT] Docs Wayland on headless no GPU server

[vincemulhollon]

Is this a new feature request?

• I have searched the existing issues

Wanted change

I'm looking for guidance on getting Wayland running on the latest ubuntu-kde webtop on a headless no GPU server.

I don't have much (well, any) experience with Wayland and simply adding the PIXELFLUX_WAYLAND true env variable isn't doing it.

I'm probably missing something obvious.

Reason for change

Ubuntu-KDE is now tagged as "Wayland Only" and no longer works on a headless no GPU server.

Proposed code change

I guess this is a feature request on https://github.com/linuxserver/docker-webtop to document how to set up a Wayland only webtop on a headless no GPU server.

[github-actions]

Thanks for opening your first issue here! Be sure to follow the relevant issue templates, or risk having this issue marked as invalid.

[thelamer]

It works fine without a GPU, it does not work on processors without AVX2 support, how old is your proc?

[thelamer]

IE:

docker run --rm -it -p 3001:3001 --shm-size=1gb linuxserver/webtop:ubuntu-kde bash

Plasma is aware it is running on pixman as well and limits the accelerated desktop features.

2026-04-10.09-56-24.mp4

[vincemulhollon]

It works fine without a GPU, it does not work on processors without AVX2 support, how old is your proc?

ha ha both new... and old.

I'd suggest adding this a "feature request" to add documentation text something like this to any container moving to, or configured to use wayland, or adding it to the existing "how to use wayland" docs:

"Wayland will crash without AVX2 CPU instruction support. Most post-2008 processors support AVX2. An example of a system that by default does not support AVX2 instructions is a Proxmox 8.0 or newer VM with the Proxmox default standard CPU ABI configuration of x86-64-v2-AES. AVX is not supported until cpu ABI x86-64-v3 or newer, so simply upgrade the VM hardware exposed CPU to v3 or perhaps 'host' passthru, or upgrade to some CPU Type ABI level that supports AVX2"

Some references:

https://pve.proxmox.com/wiki/Roadmap#Proxmox_VE_8.0

See 7th "Highlight" on the list of Highlights. Proxmox 8.0 and newer (released in summer of 2023) defaults all newly created VMs to CPU ABI type x86-64-v2-AES which makes it easy to migrate between hosts in a heterogenous cluster because everything post 2007 or so supports v2. So I can live-migrate my VMs or high availability restore from my very newest cluster host to any old potato host without issue, and it works great, cool. However...

https://gitlab.com/x86-psABIs/x86-64-ABI

On about page 16 of the cpu ABI reference pdf, v2 does not support AVX2 but v3 does support AVX and AVX2.

So, the default configuration of Proxmox as shipped, prevents Wayland from running on Proxmox clusters, which is really super annoying.

I tried a Proxmox VM CPU type of host pass-thru (NOT the default CPU ABI for proxmox and not a good idea in a heterogeneous cluster) and everything works great! Cool! No crashes... yet. I tried this on a VM hosting docker and on the K8S test cluster of VMs and they all work. Pretty cool! Its all good.

x86-64-v4 cpu ABI isn't supported until post-2022 CPUs so I think in the long run I need to change the entire Proxmox cluster from default cpu ABI x86-64-v2-AES to ABI v3. Some of my heterogenous cluster will not support v4 but I think anything made after 2008 or so will run cpu ABI v3. I can't use "host" or my VM will be "stuck" on the exact model of host that its initially configured upon... I think.

This is the first, and so far only, application I've ever hosted on the cluster that requires post x86-64-v2 CPI ABI, I guess this was bound to happen sooner or later. Very interesting.

[thelamer]

I would kick this upstream that seems like a really stupid default configuration, especially given the age of the Haswell architecture. They should at least hook and detect capabilities before setting the cpu profile.

Just in general what you see as not working is because there is no fallback specifically on the 2 kde tags (ubuntu and alpine) as they have literally ripped out the bins for it, normally it just falls back to X11 mode for all other images:

https://github.com/linuxserver/docker-baseimage-selkies/blob/master/root/etc/s6-overlay/s6-rc.d/init-selkies-config/run#L3-L8

I think my only task here is to add a x11 startwm.sh that splashes a window saying "your platform is not supported", our readme is getting really long.

[vincemulhollon]

Perhaps something like "cat /proc/cpuinfo | grep avx2" needs to find a avx2 capability.

Well at any rate, yeah, its a CPU ABI version issue that would mostly hit people running in VMs on clusters.

[thelamer]

It does do that check, look at the code I linked you, I updated the image to splash the information when it is initialized in X11 mode.