KMS + Headless + Nvidia: Unknown Monitor connector type [Meta]
#2250
Comments
|
I also experience this on Hyprland. Creating the headless monitor with |
|
I get the same error on Hyprland - Funny as my 1. setup worked with nvidia and hyprland, some updates happened (dont know which one exactly it was) and now it seems to be borked |
|
Installed hyprland and repleaced hyprland-git, headless streaming via sunshine works for me now again. |
|
I just want to note that the breaking change seems to originate from hyprwm/Hyprland@4f26ae7 The developers of Hyprland are not interested in fixing this issue as it is their opinion that a different function should be in use as wlr_dmabuf has been apparently superseded. Will this be able to be resolved within the Sunshine codebase? |
|
Anyone is free to make a PR. If you want stability though you should probably use x11. The Wayland protocol they superceded was unstable and the replacement one is also unstable. |
|
I've had linked the wrong issue, and I've corrected it now. Apologies. So there's no intention to shift to wlr_screencopy_v1? |
|
I am also in the boat of making heavy use of hyprland's easy to create headless monitor feature, streaming it using Sunshine, so I would also interested in seeing this being updated to using wlr_screencopy_v1 over wlr_export_dmabuf, although I currently lack the proper knowledge to contribute to that change myself In the meantime, for those using nix, rolling back to the previous version gets it back to a working state. You can use Lazamar's tool for getting a previous revision, if you don't use flakes, but you can also just look up the history of commits to the hyprland nix package and just get one that works, as Lazamar's only updates every now and again. |
|
Same issue here, running headless on NixOS with NVIDIA GPU. Cloudy Pad relies entirely on the ability to use virtual monitor which is seemingly broken now. If I understand correctly, this Sunshine logs is pointing to the real issue: Causing Sunshine to fallback to capture method non-compatible with virtual display. Is there a possible workaround ? |
|
Same problem with weaton wayland, chatgpt Said i should try to rebuild with wlr_screencopy, i ll try These days and Tell u output |
ref: hyprwm/Hyprland#6623 ref: LizardByte/Sunshine#2250 requires sunshine to implement support for the newer protocol that hyprland continues to expose
|
Did anyone get this back to working without downgrading hyprland version to the last commit where wlroots are available? |
@fxzzi Have you since found a workaround? This still hasn't been fixed, I've opened an issue. |
|
Have the same error when using virtual display on GNOME 47 and Wayland |
|
same problem on Fedora 40. Any updates? |
|
It seems this issue hasn't had any activity in the past 90 days. If it's still something you'd like addressed, please let us know by leaving a comment. Otherwise, to help keep our backlog tidy, we'll be closing this issue in 10 days. Thanks! |
|
Bumping this, as it's still an issue |
|
Hello, i'm a new Sunshine / Moonlight user here, only testing so far. I know nothing about coding or applying patches from github or anything, but my aim is to run headless 4K@60 desktop VM's with pcie passthrough to GPU (Quadro P620) on proxmox. Have been trying to read up, and i know the easiest solution is a dummy 4K plug, the second easiest is to do the X11 trick. But ofcourse, a headless VM running wayland with no dummy plug, and streaming via sunshine is the goal. Reason i decided to barge into this thread is i have a fully working test setup here, latest proxmox with working pcie passthrough to Quadro P620 GPU, hardware keyb/mouse access to Both client running Moonlight and VM running Sunshine, and 2x 4K@60 screens connected to client and server in front of me. I have not yet tested hyprland, it was planned, but i can test anything easy on this setup. Currently a test VM running Fedora 41 standard install w/ Wayland, pcie passthrough to P620 with a working Sunshine installed from COPR beta. I want to help, so if anyone has some ideas that i could easily test with this setup, just tell me what to do, and i can report back. |
Is there an existing issue for this?
Is your issue described in the documentation?
Is your issue present in the nightly release?
Describe the Bug
Hello!
First of all, thank you for this outstanding project.
The only way I got my headless setup to work is with the X11 capture method, unfortunately. I have been trying to get my headless setup to work with KMS to no avail. I am now trying to get Wayland + KMS working - since it should be an option since 0.22 - and I just got this "new" error:
I am creating a virtual monitor with:
Without creating the virtual monitor,
sunshinealways fails withError: Unable to initialize capture method.Modeset is enabled (
nvidia-drm.modeset=1).Expected Behavior
No response
Additional Context
[min_log_level] -- [0]
Host Operating System
Linux
Operating System Version
NixOS 23.11
Architecture
64 bit
Sunshine commit or version
0.22.0
Package
Linux - nixpkgs (Third Party)
GPU Type
Nvidia
GPU Model
RTX 4090
GPU Driver/Mesa Version
535.129.03
Capture Method (Linux Only)
KMS
Config
Apps
No response
Relevant log output
The text was updated successfully, but these errors were encountered: