discussion moved here

 

Docker container for viewing the utilization of an Intel GPU.

  

Dockerhub URL:

https://hub.docker.com/r/djaydev/intel-gpu-tools

  

If you would like to use an Unraid template, download and place the following xml file in your flash drive's config/plugins/dockerMan/templates-user folder.  Then from the unraid UI click add container and select intel-gpu-tools in the template drop down.

unraid template link:

https://raw.githubusercontent.com/djaydev/docker-intel-gpu-tools/master/unraid/template/intel-gpu-tools.xml

Edited February 1, 20197 yr by dee31797

Will you be adding this to CA?

1 hour ago, daves said:

Will you be adding this to CA?

It's in CA now, let me know if you have any issues.

Just installed.  Works great!  Is there anyway to disable the webserver/VNC server?  I really only need to run this from the command line.

17 hours ago, daves said:

Just installed.  Works great!  Is there anyway to disable the webserver/VNC server?  I really only need to run this from the command line.

There's no way to disable the webserver/VNC server but with so little resources it uses there's really no reason to do so.  If you're worried about outside access the best bet is to not assign any ports to the container, you can just delete those entries from the container config in the WebGUI.

 

If you're still set on not running the webserver/VNC servers, I recommend not bothering with prebuilt containers and just building your own locally.

Save the following text to a file called "Dockerfile"

FROM debian

RUN apt-get update && \
	apt-get install intel-gpu-tools bash -y && \
	rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/*

Change directory to where the Dockerfile is saved and run the cli command:

docker build -t myfavdockerimage .

That will create your customized docker image.  Replace "myfavdockerimage" with whatever you want the name to be.

Now whenever you want to run your image from CLI:

docker run --rm -ti --privileged=true --device /dev/dri:/dev/dri myfavdockerimage intel_gpu_tools

 

Edited May 8, 20197 yr by dee31797

Quote

 If you're worried about outside access the best bet is to not assign any ports to the container, you can just delete those entries from the container config in the WebGUI.

Done!  Works for me!

Hi there, I tried to get this tool/docker working but unfortunately iam failing everytime

 

Iam getting a black screen with these logs:

 

 

[app] starting intel-gpu-tools...

[app] starting intel-gpu-tools...
Failed to connect to session manager: Failed to connect to the session manager: SESSION_MANAGER environment variable not defined

 

9 minutes ago, Toobie said:

Hi there, I tried to get this tool/docker working but unfortunately iam failing everytime

 

Iam getting a black screen with these logs:

 

 

[app] starting intel-gpu-tools...

[app] starting intel-gpu-tools...
Failed to connect to session manager: Failed to connect to the session manager: SESSION_MANAGER environment variable not defined

 

 

That error usually shows up when the container can't access the intel iGPU.  You have to ensure the iGPU shows up in /dev/dri/

ls /dev/dri/

Might want to try https://www.reddit.com/r/unRAID/comments/9eyt8u/unraid_intel_quick_sync_setup_for_plex/

 

 

I just ran headlong into this and realized that I had at some point replaced my go and syslinux.cfg without adding back the i915 modifications that were required.

 

This thread was more useful to me at this point in time than the reddit thread (looks like some stuff was deleted from it): https://forums.unraid.net/topic/62525-solved-igpu-transcoding-i915-driver-present-but-devdri-not-available/?do=findComment&comment=703961

On 7/10/2019 at 10:27 AM, dee31797 said:

 

That error usually shows up when the container can't access the intel iGPU.  You have to ensure the iGPU shows up in /dev/dri/

ls /dev/dri/

Might want to try https://www.reddit.com/r/unRAID/comments/9eyt8u/unraid_intel_quick_sync_setup_for_plex/

 

 

I'm getting that error too, but I can see `/dev/dri` and plex is using hw transcode.

3 minutes ago, cherrybullet said:

I'm getting that error too, but I can see `/dev/dri` and plex is using hw transcode.

can you please post the complete docker settings you're using?

7 minutes ago, dee31797 said:

can you please post the complete docker settings you're using?

Just now, cherrybullet said:

 

You have the /dev/dri mapped twice.  Please remove one and retry, let me know if the error goes away or not

12 minutes ago, dee31797 said:

You have the /dev/dri mapped twice.  Please remove one and retry, let me know if the error goes away or not

I removed it from extra params and got the same error

 

Not sure if this helps but, I don't know if USER_ID=99 is correct for me. That's just a setting I saw on someone else's setup. Without it, it gives me an error saying it needs USER_ID.

7 minutes ago, cherrybullet said:

I removed it from extra params and got the same error

 

Not sure if this helps but, I don't know if USER_ID=99 is correct for me. That's just a setting I saw on someone else's setup. Without it, it gives me an error saying it needs USER_ID.

'USER_ID'='0'

4 minutes ago, dee31797 said:

'USER_ID'='0'

getting the SESSION_MANAGER error again

 

[services.d] starting app...
[app] starting intel-gpu-tools...
[services.d] done.
Failed to connect to session manager: Failed to connect to the session manager: SESSION_MANAGER environment variable not defined

 

The default template I pulled from CA had /dev/dri twice. I changed it to /dev/dri and it still didn't launch. Showing advanced settings displayed the USER_ID variable, which had no value. I added 0 and everything started working. Thanks for the container! It was interesting spying on my Plex hw transcodes.

Edited March 29, 20206 yr by Dase
Typo

57 minutes ago, Dase said:

The default template I pulled from CA had /dev/dri twice. I changed it to /dev/dri and it still didn't launch. Showing advanced settings displayed the USER_ID variable, which had no value. I added 0 and everything started working. Thanks for the container! It was interesting spying on my Plex hw transcodes.

if you're talking about "/dev/dri: /dev/dri" vs " /dev/dri" thats fine either one works.  The other user had it mapped twice, "--device  /dev/dri: /dev/dri" and another  "--device  /dev/dri: /dev/dri".

1 hour ago, cherrybullet said:

getting the SESSION_MANAGER error again

 

[services.d] starting app...
[app] starting intel-gpu-tools...
[services.d] done.
Failed to connect to session manager: Failed to connect to the session manager: SESSION_MANAGER environment variable not defined

 

cherrybullet,

Please update to the latest image and let me know if it works for you.  Docker tab -> advanced view -> force update intel-gpu-tools is my prefer method.

 

long version

For some reason this error has started occurring with the latest base image update and xfce terminal, I found that if I opened a webbrowser to the webui of the docker container and set the docker container to auto-restart on failure, it would then work on the next restart.  Obviously this shouldn't happen, but I didn't care to investigate it further so I just switched terminals.  Seems to work well on my test machine, please let me know if you still have issues.

33 minutes ago, dee31797 said:

cherrybullet,

Please update to the latest image and let me know if it works for you.  Docker tab -> advanced view -> force update intel-gpu-tools is my prefer method.

 

long version

For some reason this error has started occurring with the latest base image update and xfce terminal, I found that if I opened a webbrowser to the webui of the docker container and set the docker container to auto-restart on failure, it would then work on the next restart.  Obviously this shouldn't happen, but I didn't care to investigate it further so I just switched terminals.  Seems to work well on my test machine, please let me know if you still have issues.

That did it! Thank you for your help and taking the time to debug this

Anyone know what happened to this one? It's removed from Docker hub and github repo is gone