Hey guys,

I need your help. I'm running Unraid 7.2.4/6.12.54-Unraid on a AM5/Ryzen 7 machine with a NVIDIA GeForce RTX 5060 Ti, currently

Nvidia Driver Version: 595.58.03

Open Source Kernel Module: Yes

Since the upgrade, I'm running into unfixable erros in a steam-headless container I'm using since more than a year for moonlight based game streaming to different devices in my network. At the moment, there's a X11 related boot loop, which I can't fix with my very basic knowledge/abilities and also some research didn't help. For details, see the attached log-File.

Claude 4.6 Sonnet Thinking states the following:
"The root cause is a confirmed bug in NVIDIA driver 595.58.03 combined with the RTX 5060 Ti (Blackwell GB206) in headless containers:

•  nv_drm_atomic_commit  flip timeouts in headless mode → Xorg watchdog kills it (exit 130)

• GSP firmware polls display hotplug every ~50ms with no monitor connected → infinite DFP loop →  Deleting GPU-0 

• No fix from NVIDIA yet for this specific scenario"

Is this plausible and actually the cause? I can't downgrade to for e.g. 580.142, since without the open source option, my GPU won't be detected anymore, so I'm stuck on 595.58.03 at the moment?!

I can't figure out, if there's a work around for me nad also others experiencing a comparable problem. Any help is very much appreciated! Thx in advance!

Kind regards,

Chris

Xorg.55.log

steam-headless.log

diagnostics-20260426-1629.zip

Edited April 26Apr 26 by anticholium

16 hours ago, anticholium said:

I can't figure out, if there's a work around for me nad also others experiencing a comparable problem. Any help is very much appreciated! Thx in advance!

First of all you have a lot of those lines in your syslog:

Apr 26 15:38:31 Futro kernel: input: Mouse passthrough as /devices/virtual/input/input5

Apr 26 15:38:31 Futro kernel: input: Mouse passthrough (absolute) as /devices/virtual/input/input6

Apr 26 15:38:31 Futro kernel: input: Keyboard passthrough as /devices/virtual/input/input7

Apr 26 15:38:31 Futro kernel: input: Sunshine Nintendo (virtual) pad as /devices/virtual/input/input8

Apr 26 15:38:31 Futro kernel: input: Sunshine X-Box One (virtual) pad as /devices/virtual/input/input9

Apr 26 15:38:31 Futro kernel: nvidia-modeset: WARNING: GPU:0: Unable to read EDID for display device HDMI-0

### [PREVIOUS LINE REPEATED 1 TIMES] ###

Apr 26 15:38:34 Futro acpid: input device has been disconnected, fd 11

Apr 26 15:38:35 Futro kernel: nvidia-modeset: WARNING: GPU:0: Unable to read EDID for display device HDMI-0

### [PREVIOUS LINE REPEATED 2 TIMES] ###

Apr 26 15:38:36 Futro kernel: traps: light-locker[74379] trap int3 ip:14dde23fea77 sp:7fff1a91b290 error:0 in libglib-2.0.so.0.8400.4[66a77,14dde23b7000+a1000]

Apr 26 15:38:36 Futro kernel: nvidia-modeset: WARNING: GPU:0: Unable to read EDID for display device HDMI-0

### [PREVIOUS LINE REPEATED 3 TIMES] ###

Not sure what causes that since I'm not very familiar with the Steam Headless container.

Please also post your docker run command if possible.

Didn't you also have to create a modprobe.d file for the container to properly run?

16 hours ago, anticholium said:

Is this plausible and actually the cause? I can't downgrade to for e.g. 580.142, since without the open source option, my GPU won't be detected anymore, so I'm stuck on 595.58.03 at the moment?!

Yes, that's caused because 50 series needs to have the OpenSource driver installed.

11 hours ago, ich777 said:

Yes, that's caused because 50 series needs to have the OpenSource driver installed.

So is it not possible to go back to version 590 of the open-source driver when using a 50-series card? I have driver "590.48.01" listed as a supported candidate, but it does not recognize my 5060 Ti. Unfortunately, 595.58.03 has caused many issues for me.

1 hour ago, 123marvin123 said:

So is it not possible to go back to version 590 of the open-source driver when using a 50-series card? I have driver "590.48.01" listed as a supported candidate, but it does not recognize my 5060 Ti. Unfortunately, 595.58.03 has caused many issues for me.

Do you have Diagnostics or anything?

Please post your Diagnostics if something is not working.

On 4/27/2026 at 8:26 AM, ich777 said:

First of all you have a lot of those lines in your syslog:

Apr 26 15:38:31 Futro kernel: input: Mouse passthrough as /devices/virtual/input/input5

Apr 26 15:38:31 Futro kernel: input: Mouse passthrough (absolute) as /devices/virtual/input/input6

Apr 26 15:38:31 Futro kernel: input: Keyboard passthrough as /devices/virtual/input/input7

Apr 26 15:38:31 Futro kernel: input: Sunshine Nintendo (virtual) pad as /devices/virtual/input/input8

Apr 26 15:38:31 Futro kernel: input: Sunshine X-Box One (virtual) pad as /devices/virtual/input/input9

Apr 26 15:38:31 Futro kernel: nvidia-modeset: WARNING: GPU:0: Unable to read EDID for display device HDMI-0

### [PREVIOUS LINE REPEATED 1 TIMES] ###

Apr 26 15:38:34 Futro acpid: input device has been disconnected, fd 11

Apr 26 15:38:35 Futro kernel: nvidia-modeset: WARNING: GPU:0: Unable to read EDID for display device HDMI-0

### [PREVIOUS LINE REPEATED 2 TIMES] ###

Apr 26 15:38:36 Futro kernel: traps: light-locker[74379] trap int3 ip:14dde23fea77 sp:7fff1a91b290 error:0 in libglib-2.0.so.0.8400.4[66a77,14dde23b7000+a1000]

Apr 26 15:38:36 Futro kernel: nvidia-modeset: WARNING: GPU:0: Unable to read EDID for display device HDMI-0

### [PREVIOUS LINE REPEATED 3 TIMES] ###

Not sure what causes that since I'm not very familiar with the Steam Headless container.

Please also post your docker run command if possible.

Didn't you also have to create a modprobe.d file for the container to properly run?

Yes, that's caused because 50 series needs to have the OpenSource driver installed.

I also have no idea what exactly is causing these lines; it might be related to the steam-headless container, since it simulates a variety of input devices!? But that’s just my more or less clueless guess. – There must have been some change related to the latest driver, though, because the container was working perfectly fine just a week earlier without any changes to the configuration. – My sons use it to stream Steam games via Moonlight to the Apple TV and control them with a Nintendo Pro Controller—it’s actually pretty cool, better than some dedicated gaming consoles. But now, of course, they’re really sad because I can’t get it working again…

My docker run command:

docker run
  -d
  --name='steam-headless'
  --net='host'
  --pids-limit 2048
  -e TZ="Europe/Berlin"
  -e HOST_OS="Unraid"
  -e HOST_HOSTNAME="Futro"
  -e HOST_CONTAINERNAME="steam-headless"
  -e 'USER_PASSWORD'='password'
  -e 'TZ'='Europe/Frankfurt'
  -e 'USER_LOCALES'='en_US.UTF-8 UTF-8'
  -e 'NVIDIA_PATCH_VERSION'=''
  -e 'WEB_UI_MODE'='vnc'
  -e 'NVIDIA_VISIBLE_DEVICES'='all'
  -e 'NVIDIA_DRIVER_CAPABILITIES'='all'
  -e 'DISPLAY'=':55'
  -e 'MODE'='primary'
  -e 'PORT_NOVNC_WEB'='8083'
  -e 'ENABLE_VNC_AUDIO'='false'
  -e 'ENABLE_EVDEV_INPUTS'='true'
  -e 'CA_TS_FALLBACK_DIR'='/home/default'
  -l net.unraid.docker.managed=dockerman
  -l net.unraid.docker.webui='http://[IP]:[PORT:8083]/'
  -l net.unraid.docker.icon='https://raw.githubusercontent.com/Josh5/docker-steam-headless/master/images/steam-icon.png'
  -v '/mnt/user/appdata/steam-headless':'/home/default':'rw'
  -v '/mnt/user/Games':'/mnt/games':'rw'
  --hostname='SteamHeadless'
  --add-host='SteamHeadless:127.0.0.1'
  --restart='unless-stopped'
  --shm-size='2G'
  --ipc='host'
  -v '/tmp/.X11-unix/':'/tmp/.X11-unix/':'rw'
  -v '/tmp/pulse/':'/tmp/pulse/':'rw'
  --ulimit='nofile=1024:524288'
  --device='/dev/fuse'
  --device='/dev/uinput'
  --device-cgroup-rule='c 13:* rmw'
  --cap-add='NET_ADMIN'
  --cap-add='SYS_ADMIN'
  --cap-add='SYS_NICE'
  --security-opt='seccomp=unconfined'
  --runtime='nvidia' 'josh5/steam-headless:latest'

205d87be9c5d0aeca4d6771154e9f9cf36023455fbd2e6ccefd16f4ed94d4481

Edited April 28Apr 28 by anticholium

2 minutes ago, anticholium said:

  -e 'NVIDIA_VISIBLE_DEVICES'='all'

I would strongly recommend that you change that to you actual GPU UUID.

Just now, ich777 said:

I would strongly recommend that you change that to you actual GPU UUID.

I already tried that, no difference

Just now, anticholium said:

I already tried that, no difference

An what about removing the --runtime nvidia and these two variables and adding --gpus all

2 minutes ago, ich777 said:

An what about removing the --runtime nvidia and these two variables and adding --gpus all

Sorry, which two variables? --runtime nvidia and what?

Edited April 28Apr 28 by anticholium

1 minute ago, anticholium said:

Sorry, which to variables? --runtime nvidia and what?

Remove the two NVIDIA_* variables and replace the extra parameter --runtime nvidia with --gpus all

got it re-"running"(i.e. crashlooping) aigain, no matter which nvidia variable I ex-/include, no difference

I tried this one here:

docker run --rm --runtime=nvidia --gpus all nvidia/cuda:13.2.0-base-ubuntu24.04 nvidia-smi
Unable to find image 'nvidia/cuda:13.2.0-base-ubuntu24.04' locally
13.2.0-base-ubuntu24.04: Pulling from nvidia/cuda
01d7766a2e4a: Pull complete 
52ab058d967b: Pull complete 
33dea4171410: Pull complete 
4f4fb700ef54: Pull complete 
2761e0c7dc1e: Pull complete 
314a2f652b6c: Pull complete 
Digest: sha256:36cccda4bebc3b0b1ebe1907ead8169cf144d45df890be871b36b304cf91145a
Status: Downloaded newer image for nvidia/cuda:13.2.0-base-ubuntu24.04
ERROR: driverInitFileInfo 578 result=11ERROR: init 664 result=11ERROR: init 250 result=11Tue Apr 28 19:04:49 2026       
+-----------------------------------------------------------------------------------------+
| NVIDIA-SMI 595.58.03              Driver Version: 595.58.03      CUDA Version: 13.2     |
+-----------------------------------------+------------------------+----------------------+
| GPU  Name                 Persistence-M | Bus-Id          Disp.A | Volatile Uncorr. ECC |
| Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
|                                         |                        |               MIG M. |
|=========================================+========================+======================|
|   0  NVIDIA GeForce RTX 5060 Ti     Off |   00000000:01:00.0 Off |                  N/A |
|  0%   31C    P8              6W /  180W |       5MiB /  16311MiB |      0%      Default |
|                                         |                        |                  N/A |
+-----------------------------------------+------------------------+----------------------+

+-----------------------------------------------------------------------------------------+
| Processes:                                                                              |
|  GPU   GI   CI              PID   Type   Process name                        GPU Memory |
|        ID   ID                                                               Usage      |
|=========================================================================================|
|  No running processes found                                                             |
+-----------------------------------------------------------------------------------------+

result=11 fault = same as steam-headless... Any ideas about this?

Edited April 28Apr 28 by anticholium

35 minutes ago, anticholium said:

I tried this one here:

And if you redownload the driver, does it say the checksum matches?

37 minutes ago, ich777 said:

And if you redownload the driver, does it say the checksum matches?

I redownloaded the driver twice during the last two days, but I didn't realize any checksum mismatch and I think that would have been somewaht obvious?

I tried if Xorg is able to initialize a screen on the unraid host:

cat > /tmp/xorg-nvidia-test.conf <<'EOF'
Section "ServerLayout"
    Identifier "Layout0"
    Screen 0 "Screen0"
EndSection

Section "Monitor"
    Identifier "Monitor0"
EndSection

Section "Device"
    Identifier "Device0"
    Driver "nvidia"
    BusID "PCI:1:0:0"
    Option "AllowEmptyInitialConfiguration" "true"
    Option "UseDisplayDevice" "DFP-0"
EndSection

Section "Screen"
    Identifier "Screen0"
    Device "Device0"
    Monitor "Monitor0"
    DefaultDepth 24
    SubSection "Display"
        Depth 24
        Modes "1920x1080" "1600x900"
        Virtual 1920 1080
    EndSubSection
EndSection

Section "ServerFlags"
    Option "AutoAddGPU" "false"
EndSection
EOF

/usr/libexec/Xorg -config /tmp/xorg-nvidia-test.conf -logfile /tmp/Xorg-nvidia-test.log -noreset -novtswitch -sharevts -nolisten tcp -verbose vt7 :77

Result:

NVIDIA(GPU-0): Failed to acquire modesetting permission.
NVIDIA(0): Failing initialization of X screen
no screens found

Host nvidia-smi: OK

nvidia-container-cli: OK

CUDA container nvidia-smi: OK

/dev/dri/card0: korrekt RTX 5060 Ti

nvidia_drm.modeset: Y

HDMI-Display an RTX 5060 Ti: connected

Container-Xorg Scanout: FAIL

Host-Xorg Scanout: FAIL

Container-Xorg NoScanout: OK

Sunshine mit NoScanout: FAIL, weil 0 Outputs

So maybe the culprit might be the Unraid/NVIDIA-595.58.03/Open-Kernel/RTX-5060-Ti-Host-Stack??

Edited April 28Apr 28 by anticholium

6 minutes ago, anticholium said:

Xorg crashes even with a connected & detected real hdmi display :-(

Please report that in the Steam Headless container.
Also other users where reporting that the Steam container from Linuxserver.io was working just fine.

I can't tell you why the container is crashing, in terms of the driver everything seems perfectly fine and I think now that this isn't a driver issue anymore.

And by the love of god, can you please hide such long posts behind a spoiler and avoid double and triple posting, this is not a chat like Discord.

You are completely bloating this thread.

EDIT:

2 minutes ago, anticholium said:

So maybe the culprit might be the Unraid/NVIDIA-595.58.03/Open-Kernel/RTX-5060-Ti-Host-Stack??

No, I don't think so because it seems that not many users have that issue like you have and the Driver is for all cards the same.

2 minutes ago, ich777 said:

Please report that in the Steam Headless container.
Also other users where reporting that the Steam container from Linuxserver.io was working just fine.

I can't tell you why the container is crashing, in terms of the driver everything seems perfectly fine and I think now that this isn't a driver issue anymore.

And by the love of god, can you please hide such long posts behind a spoiler and avoid double and triple posting, this is not a chat like Discord.

You are completely bloating this thread.

Ok, sorry for the mess, I can remove the postings. But what about Xorg failing to initialize a screen on the host itself?

Sadly, the linuxserver.io-image is not moonlight capable, "just" web-accessible. Furthermore: it only supports 580+ full proprietary drivers, which is not an option for my blackwell gpu...

Edited April 28Apr 28 by anticholium

2 minutes ago, anticholium said:

Ok, sorry for the mess, I can remove the postings. But what about Xorg failing to initialize on the host itself?

Sorry but do you boot Unraid in GUI mode?
As far as I can see you are just trying to run Xorg without anything configured, did your run nvidia-xconfig or anything like that?
Did you configure Xorg to use your Nvidia card? Did you configure a Window manager?

I think you are getting that wrong how that works, if not everything is installed on the host (in terms of Xorg and dependencies) than it won't work properly and cause issues.

However did you yet try the Steam container from Linuxserver.io?

As said above it seems like you are the only one who has currently an issue like that, sure there where issues because I upgraded the container runtime and made a mistake in that process, but that is now all fixed.

3 minutes ago, ich777 said:

Sorry but do you boot Unraid in GUI mode?
As far as I can see you are just trying to run Xorg without anything configured, did your run nvidia-xconfig or anything like that?
Did you configure Xorg to use your Nvidia card? Did you configure a Window manager?

I think you are getting that wrong how that works, if not everything is installed on the host (in terms of Xorg and dependencies) than it won't work properly and cause issues.

However did you yet try the Steam container from Linuxserver.io?

As said above it seems like you are the only one who has currently an issue like that, sure there where issues because I upgraded the container runtime and made a mistake in that process, but that is now all fixed.

No, I'm not running Unraid in GUI mode. I configured Xorg:

cat > /tmp/xorg-nvidia-test.conf <<'EOF'

Section "ServerLayout"

Identifier "Layout0"

Screen 0 "Screen0"

EndSection

Section "Monitor"

Identifier "Monitor0"

EndSection

Section "Device"

Identifier "Device0"

Driver "nvidia"

BusID "PCI:1:0:0"

Option "AllowEmptyInitialConfiguration" "true"

Option "UseDisplayDevice" "DFP-0"

EndSection

Section "Screen"

Identifier "Screen0"

Device "Device0"

Monitor "Monitor0"

DefaultDepth 24

SubSection "Display"

Depth 24

Modes "1920x1080" "1600x900"

Virtual 1920 1080

EndSubSection

EndSection

Section "ServerFlags"

Option "AutoAddGPU" "false"

EndSection

EOF

/usr/libexec/Xorg -config /tmp/xorg-nvidia-test.conf -logfile /tmp/Xorg-nvidia-test.log -noreset -novtswitch -sharevts -nolisten tcp -verbose vt7 :77

Sadly the linuxserver.io image isn't a moonlight compatible replacement...

Edited April 28Apr 28 by anticholium

2 minutes ago, anticholium said:

No, I'm not running Unraid in GUI mode. I configured Xorg:

But wouldn't it be way easier to run Unraid in GUI mode since this would also start Xorg...?

However you still didn't answer, have you yet tried the Steam container from Linxuserver.io?
Do you have other container which are using the Nvidia driver, if yes do this containers work, like for example transcoding container, Plex, Emby, Jellyfin, Unmanic, Tadarr,...?

If one of the containers is working then it's not the driver who's at fault usually.

I also don't like the config:

Option "UseDisplayDevice" "DFP-0"

Modes "1920x1080" "1600x900"

Virtual 1920 1080

since these can cause trouble.

This is not the correct way to test things.

I didn't try the linuxserver.io image, since it doesn't have moonlight/sunshine support implemented, it's "only" web accessible... – Maybe I should try it nontheless. – Concerning the xorg test on the host itself: how would a window manager or a desktop session help without an initialized/initializable display? I thougt its roughly this: xorg starts, loads the driver, gpu/display/screen is initialized, x display is provided, then the window manager!? No? But the host xorg logs show that it's already failing at step 3, initializing the display, like in the steam-headless container. I just thought, that even a full blown desktop environment would fail also with high probability because of that? But I will try to run unraid in GUI mode and run the test again. Thank you for your efforts, I really appreciate your answers!

from the logs:
NVIDIA dlloader X Driver 595.58.03

NVIDIA GLX Module 595.58.03

NVIDIA GPU NVIDIA GeForce RTX 5060 Ti (GB206-A) at PCI:1:0:0

Option "UseDisplayDevice" "DFP-0"
...
NVIDIA(GPU-0): Failed to acquire modesetting permission.

NVIDIA(0): Failing initialization of X screen

PS: yes, Plex for eg successfully uses the gpu for transcoding
PPS: linuxserver.io: "Nvidia support only works on 580 and up full proprietary drivers (no MIT/GPL)" so not suitable for me at all (blackwell architecture, open source, you know that way better than me).

Edited April 28Apr 28 by anticholium

another unraid user was able to resolve the same bug by rolling back to 580.142, since he has a RTX 3060, so not dependent on open source kernel modules. 590.48.01 did only work partially... Any thougts on that? the linuxserver image, as expected, didn't work because of not supporting open source driver... – any "clean" option to manually downgrade to https://github.com/unraid/unraid-nvidia-driver/releases/download/6.12.54-Unraid/nvos-580.95.05-6.12.54-Unraid-1.txz ?

https://github.com/Steam-Headless/docker-steam-headless/issues/224

Edited April 30Apr 30 by anticholium

On 4/28/2026 at 9:43 PM, ich777 said:

And if you redownload the driver, does it say the checksum matches?

59 minutes ago, anticholium said:

https://github.com/Steam-Headless/docker-steam-headless/issues/224

Sorry but this was 28 days ago and with the old Nvidia Driver and the driver was compiled 5 days ago:

30 minutes ago, anticholium said:

That's the correct checksum yes:

https://github.com/unraid/unraid-nvidia-driver/releases/download/6.12.54-Unraid/nvos-595.58.03-6.12.54-Unraid-1.txz.md5

I don't know what to do else sorry.

Oh, that's weird, I obviously didn't realize that before. Thank you nontheless.