Open shyim opened 4 years ago
I've also just discovered that we can choose different nvidia packages. Besides that I think I'm out of things to try.
I tried a mostly plain configuration as well as nvidia_x11
and nvidia_x11_beta
. Thanks for trying, I'm going to ask on discourse.
Issues with the same problem (at least I'm not alone) https://github.com/NixOS/nixpkgs/issues/85866 https://github.com/NixOS/nixpkgs/issues/66755 https://github.com/NixOS/nixpkgs/issues/132045 https://github.com/NixOS/nixpkgs/issues/82959 https://github.com/NixOS/nixpkgs/issues/94958 https://github.com/NixOS/nixpkgs/issues/101202 https://github.com/NixOS/nixpkgs/issues/55191 https://github.com/NixOS/nixpkgs/issues/84692
Hopefully they can see your solution to the issue from the linkback if anything. It might work for some people even if it doesn't work on my end.
Is there any way to escalate this?
@jonringer (nvidia-x11) @primeos @vcunat (mesa) any thoughts on how this might be resolved?
(Note: I am forwarding X11. I have tried with ssh -X, ssh -Y, ssh -XY, and vglconnect, all resulting in the above error)
X11 forwarding with OpenGL? I don't expect that's even meant to work.
The nvidia package doesn't contain a lib/dri/swrast_dri.so failed
, the only way I can think of handling this would be to add mesa to hardware.opengl.extraPackages
, so that you inherit some of the missing directories like /lib/dri
Thanks for responding.
@vcunat virtualgl (via vglconnect
and vglrun
) is supposed to handle 3D rendering on the server, and then the 2D stuff is rendered on the clients forwarded X server.
@jonringer I did that, but I still have the same error. (hardware.opengl.extraPackages = [ pkgs.mesa ];
)
Mesa doesn't appear to have that library:
find /nix/store/5j7k2nrgvl6vnichkg4c1fwj1907nvgg-mesa-21.3.5/
/nix/store/5j7k2nrgvl6vnichkg4c1fwj1907nvgg-mesa-21.3.5/
/nix/store/5j7k2nrgvl6vnichkg4c1fwj1907nvgg-mesa-21.3.5/share
/nix/store/5j7k2nrgvl6vnichkg4c1fwj1907nvgg-mesa-21.3.5/share/drirc.d
/nix/store/5j7k2nrgvl6vnichkg4c1fwj1907nvgg-mesa-21.3.5/share/drirc.d/00-mesa-defaults.conf
/nix/store/5j7k2nrgvl6vnichkg4c1fwj1907nvgg-mesa-21.3.5/lib
/nix/store/5j7k2nrgvl6vnichkg4c1fwj1907nvgg-mesa-21.3.5/lib/libglapi.so
/nix/store/5j7k2nrgvl6vnichkg4c1fwj1907nvgg-mesa-21.3.5/lib/libglapi.so.0.0.0
/nix/store/5j7k2nrgvl6vnichkg4c1fwj1907nvgg-mesa-21.3.5/lib/libgbm.so.1
/nix/store/5j7k2nrgvl6vnichkg4c1fwj1907nvgg-mesa-21.3.5/lib/libgbm.so
/nix/store/5j7k2nrgvl6vnichkg4c1fwj1907nvgg-mesa-21.3.5/lib/libgbm.so.1.0.0
/nix/store/5j7k2nrgvl6vnichkg4c1fwj1907nvgg-mesa-21.3.5/lib/libglapi.so.0
/nix/store/5j7k2nrgvl6vnichkg4c1fwj1907nvgg-mesa-21.3.5/etc
Anything you would recommend to find the missing library?
oh sorry, it's pkgs.mesa.drivers
Thanks! That resolved the error when I simply run glxgears
on my server.
Do you have any insights into why I might get the "fbconfig" error when I run it in vglrun
?
$ TVNC_WM=/run/current-system/sw/bin/twm vncserver
Desktop 'TurboVNC: host:2 (remote)' started on display host:2
Starting applications specified in /home/user/.vnc/xstartup.turbovnc
Log file is /home/user/.vnc/host:2.log
$ cat /home/user/.vnc/host:2.log | grep running
18/02/2022 03:15:45 VNC extension running!
$ vglrun -d :2 glxgears
[VGL] NOTICE: Automatically setting VGL_CLIENT environment variable to
[VGL] 192.168.1.12, the IP address of your SSH client.
Error: couldn't get an RGB, Double-buffered visual
Seems turbovnc
has the drivers linked already
nix-store -q --references $(which vncviewer) | grep drivers
/nix/store/2gl456ql1g1pj9qwkgkpjajknc1j702h-mesa-21.3.5-drivers
However virtualgl
doesn't
$ nix-store -q --references $(which vglrun) | xargs nix-store -q --references | uniq
/nix/store/4s21k8k7p1mfik0b33r2spq5hq7774k1-glibc-2.33-108
/nix/store/6pdb0rz2w2hsfsy66kwrbz67jd424wv7-libX11-1.7.2
/nix/store/a3yw11i2iskvr1zx2za4mwrj9sawq7hz-libXext-1.3.4
/nix/store/5lzlfgrc37wqj400b4qgpaa6kc9d2wr1-libglvnd-1.4.0
/nix/store/ndnqiz3nnifj1blhg9q626xlmkqq1nmh-gcc-10.3.0-lib
/nix/store/04yj0nllw5bqzzlay5m6zzbmyw7csgnk-glu-9.0.2
/nix/store/a6i6c7hkfn5s9bbgc1x6j1cnd5dppzkv-libjpeg-turbo-2.1.2
/nix/store/pi2h6nky70gww75w05zq1ch9fmycrvvy-libGL-1.4.0
/nix/store/0yb5b9wq3y7wn5l9akcz7hcyyk2bc6rk-fltk-1.3.8
/nix/store/1dw67iap73klp5rqcvqpf8sadrvr8y8d-ocl-icd-2.3.1
/nix/store/n8k71gd1a759s9mmi85iinfdb0ckkzkv-glibc-2.33-108
/nix/store/fp08p7c881knk9h135vj1acb47q218id-gcc-10.3.0-lib
/nix/store/1l82d038523bapkkc0qkgjyzdblfxbri-gcc-10.3.0
/nix/store/290xnrwys55hbnv9xggcc98q2sfgqf6g-libXtst-1.2.3
/nix/store/2dv93bbc06c7zg866qid73j3r36zz3jx-gcc-10.3.0
/nix/store/4k40gdgf7ckimb3hlwywx76dj5435ack-libXv-1.0.11
/nix/store/cmq9rapgxx18h99r0yp6vvzk1fkwjs3j-libX11-1.7.2
/nix/store/nca9ls2cla27qal0p01k2g17agmd62bf-libXext-1.3.4
/nix/store/64hrj9wbf9yvqwkjwr2r7qsdl99f8wcm-libXtst-1.2.3
/nix/store/hrc96ak11li2czfqj0xf2fc1clpn6gvx-libglvnd-1.4.0
/nix/store/aqj8zmgxvx44r5kf7jnqyyiwwfpbhj0x-libGL-1.4.0
/nix/store/bqkimn04bkknc2ngl01ipmjlq782d7m6-libjpeg-turbo-2.1.2
/nix/store/zhh1ppg81ahlvbf7qlfirj1pzlm8gizr-glu-9.0.2
/nix/store/aa5m1mpvsamdx7pi92fbkfhwh19q62z3-fltk-1.3.8
/nix/store/rycnmssyan6nd1dzas6v6b886qvk663m-ocl-icd-2.3.1
/nix/store/vnm9fwc6hj2igr5as9x2bjwx5nm8g0w8-libXv-1.0.11
/nix/store/nicc84inpa08zi5m52l30nfbp411apnw-virtualgl-lib-2.6.5
/nix/store/sp7fy3jh5ikc2f0g61h164nn3vvcky4a-virtualgl-lib-2.6.5
nix-store -q --references /run/opengl-driver
/nix/store/2gl456ql1g1pj9qwkgkpjajknc1j702h-mesa-21.3.5-drivers
/nix/store/vlp25l161q65qnm8my6zj9i87gkfh748-nvidia-x11-510.47.03-5.16.8
For now I'm trying to recreate the virtualgl package definition with pkgs.mesa.drivers
included.
export LD_LIBRARY_PATH="/run/opengl-driver/lib:/run/opengl-driver-32/lib"
resulted in vglrun -d :2 glxgears
rendering a black screen rather than an error, so it seems I have an issue that is likely unrelated now. Thanks so much for getting me past that hurdle!
That's not how finding OpenGL is supposed to work in NixOS (but we did use it in past). Nowadays it should be indirected through libglvnd. (well, I have no idea about virtualgl in particular...)
FYI: I use nixos-unstable
as my system flake. Stuff works fine when running, e.g., nix run nixos-unstable#glxinfo
. However, if I use an older flake such as nix run nixos-21.11#glxinfo
, I get the same errors mentioned in this thread.
I bumped into this because nixos-unstable#calibre
is broken at the moment, and I cannot run nixos-21-.11#calibre
either.
I know it's a tangential topic, but as for workarounds for such cases, I'd:
nix build /nix/store/70mp9j7lxbyy0ckg22y3g865y9mszlc0-calibre-5.37.0
Thank you for tthis tip! I can build calibre
, but I get the same GLX error.Failed to finding matching FBConfig
.
For reference, the above calibre starts up without any errors on my 21.11 system.
Indeed, that's interesting...
The issue is back for kitty as well.
One year later, sadly still a problem. I've hit it with alacritty and realsense-viewer. Previously, multiple times, nixGL solved it for alacritty for me, but have issues with that currently. And as for realsense-viewer, nixGL doesn't work on arm which is where I'm using the realsense-viewer.
One year later, sadly still a problem. I've hit it with alacritty and realsense-viewer. Previously, multiple times, nixGL solved it for alacritty for me, but have issues with that currently. And as for realsense-viewer, nixGL doesn't work on arm which is where I'm using the realsense-viewer.
I just got used to rebooting after upgrading, and it's not always needed.
I just got used to rebooting after upgrading, and it's not always needed.
Rebooting doesn't work in my case.
If anyone knows where the error is being thrown (source code), I'm be interested in digging into the problem.
For reproducability, I've also uploaded a 32Gb bootable image for a raspberry pi 3b here username: csce635
password bumpnrun
(obviously there's no sensitive information on it)
I'm encountering this exact same error almost 4 years later.....
I'm encountering this exact same error almost 4 years later.....
I followed the rabbit hole on this thread and solved the problem as so:
The issue was an out of date home-manager flake in ~/.config/home-manager
nix flake update home-manager switch
Boom, problem solved!
Reproduced on nixpkgs 1a9df4f74273f90d04e621e8516777efcec2802a
, on Fedora 40. Neither Kitty nor Alacrity work.
On Fedora it will be more like topics around #9415
Describe the bug Kitty does not start more
To Reproduce Steps to reproduce the behavior:
Expected behavior Its working
Screenshots
Additional context Add any other context about the problem here.
Metadata
"x86_64-linux"
Linux 5.5.5, NixOS, 20.09pre214390.ea79a830dcf (Nightingale)
yes
yes
nix-env (Nix) 2.3.3
"nixos-20.03pre208171.2e8fc97dbfa"
"nixos-20.09pre214390.ea79a830dcf"
/nix/var/nix/profiles/per-user/root/channels/nixos
Maintainer information: