[Plugin] Linuxserver.io - Unraid Nvidia


Recommended Posts

2 minutes ago, lovingHDTV said:

Hints on how to recover my previous working working version?

Chances are that you just need to rewrite all the bz* type files on the USB stick.    You can do this by downloading the zip file for a release (including rc4) and extracting the bz* files.   There should also be the files from the previous release in the ‘previous’ folder that can be copied to the root of the flash drive.

Link to comment

Appreciate all the work you guys have put in but I am stumped. I have been unable to build or update any containers once I install RC3 or RC4.  If I switch back to the standard build, no issues.  I even tried switching the older K2200 card out with a 1050 Ti.   

 

I tried multiple different containers, but all spit out errors.   Any insights? 

 

root@localhost:# /usr/local/emhttp/plugins/dynamix.docker.manager/scripts/docker run -d --name='binhex-plexpass' --net='host' --log-opt max-size='50m' --log-opt max-file='1' -e TZ="America/New_York" -e HOST_OS="Unraid" -e 'TRANS_DIR'='/config/transcode' -e 'UMASK'='000' -e 'PUID'='99' -e 'PGID'='100' -e 'NVIDIA_DRIVER_CAPABILITIES'='all' -e 'NVIDIA_VISIBLE_DEVICES'='GPU-c8814be1-3d0f-15fa-c1c4-bf0f4aa47348' -v '/mnt/user':'/media':'rw' -v '/mnt/cache/appdata/binhex-plexpass':'/config':'rw' --runtime=nvidia 'binhex/arch-plexpass' 
panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0xb2 pc=0x8b00ec]

goroutine 1 [running]:
github.com/docker/cli/vendor/github.com/spf13/pflag.(*FlagSet).VarPF(0xc0000c6800, 0x2197420, 0xc00021d1c8, 0x1e4316a, 0x6, 0x0, 0x0, 0x1e71894, 0x1f, 0x1e7622f)
/go/src/github.com/docker/cli/vendor/github.com/spf13/pflag/flag.go:817 +0x6c
github.com/docker/cli/vendor/github.com/spf13/pflag.(*FlagSet).VarP(...)
/go/src/github.com/docker/cli/vendor/github.com/spf13/pflag/flag.go:825
github.com/docker/cli/vendor/github.com/spf13/pflag.(*FlagSet).StringVar(0xc0000c6800, 0xc00021d1c8, 0x1e4316a, 0x6, 0xc0000f2da0, 0xd, 0x1e71894, 0x1f)
/go/src/github.com/docker/cli/vendor/github.com/spf13/pflag/string.go:37 +0x9d
github.com/docker/cli/cli.setupCommonRootCommand(0xc0003b4f00, 0x1c448a0, 0xbd, 0xbe)
/go/src/github.com/docker/cli/cli/cobra.go:24 +0xe1
github.com/docker/cli/cli.SetupRootCommand(0xc0003b4f00, 0xc00002ebd0, 0xc0001b1df0, 0x2)
/go/src/github.com/docker/cli/cli/cobra.go:55 +0x2f
main.newDockerCommand(0xc0001540f0, 0x11)
/go/src/github.com/docker/cli/cmd/docker/docker.go:54 +0x233
main.runDocker(0xc0001540f0, 0x2150ba0, 0xc00000e020)
/go/src/github.com/docker/cli/cmd/docker/docker.go:250 +0x40
main.main()
/go/src/github.com/docker/cli/cmd/docker/docker.go:292 +0xf1

The command failed.

 

Link to comment

I don't suppose you all can try these drivers next release, could ya?

https://www.nvidia.com/Download/driverResults.aspx/150803/en-us

Reason I'm wanting to try that driver is because money is tight right now and I can't afford to upgrade my GTX 760 I used for Transcoding before it quit working and the release notes for that version has this in it.

Fixed a regression introduced in the 430.* series of releases that caused a segmentation fault in libnvcuvid.so while using Video Codec SDK APIs on certain graphics boards.

 

Link to comment
5 hours ago, AnnabellaRenee87 said:

I don't suppose you all can try these drivers next release, could ya?

https://www.nvidia.com/Download/driverResults.aspx/150803/en-us

Reason I'm wanting to try that driver is because money is tight right now and I can't afford to upgrade my GTX 760 I used for Transcoding before it quit working and the release notes for that version has this in it.


Fixed a regression introduced in the 430.* series of releases that caused a segmentation fault in libnvcuvid.so while using Video Codec SDK APIs on certain graphics boards.

 

For the time being, we only use the current latest driver that's available at build time.

Link to comment

My Quadro stopped showing up in the plugin settings page a few weeks back, and yesterday Plex and Jellyfin both wouldn't launch after updates. I tried re-installing the plugin but it has the same issue...

 

nvidia-smi sees the card there (The 1070 is isolated for a VM, and I don't think the settings page ever saw it before).

 

Is my card dead or something??

 

 

 

 

Screenshot from 2019-10-28 14-20-52.png

Screenshot from 2019-10-28 14-21-12.png

gunhaver-diagnostics-20191028-2021.zip

Link to comment

Hey all,

Taking my first stab at installing a P2000 in my rig and it's giving me all sorts of fuss. I'm currently running 6.8rc4.

Went through the plugin and installed the nvidia version of rc4, rebooted prior to installed the GPU and was getting kernel panic errors.

 

end kernel panic - not syncing: VFS: unable to mount root fs on unknown-block(0,0)

 

I was able to revert back and get booted, so I tried the approach of installing the GPU, then going through the update processes, same troubles.

I reverted back to 6.8rc3 and rc1 and the issue persists. I'm not sure why it's giving me such an issue, I'm not terribly well versed in the Linux side of things but if anyone has some recommendations on what to try I'd appreciate it.

Link to comment
3 hours ago, m0ngr31 said:

My Quadro stopped showing up in the plugin settings page a few weeks back, and yesterday Plex and Jellyfin both wouldn't launch after updates. I tried re-installing the plugin but it has the same issue...

 

nvidia-smi sees the card there (The 1070 is isolated for a VM, and I don't think the settings page ever saw it before).

 

Is my card dead or something??

 

 

 

 

Screenshot from 2019-10-28 14-20-52.png

Screenshot from 2019-10-28 14-21-12.png

gunhaver-diagnostics-20191028-2021.zip 127.91 kB · 0 downloads

It could be that your Plex and jellyfin dockers are seeing the 1070, then failing because your vm is using them.  Try disabling the vm and seeing if it launches.  If it does, and then uses the 1070, then you know the problem.

Link to comment
On 10/27/2019 at 8:02 PM, dewlite said:

Appreciate all the work you guys have put in but I am stumped. I have been unable to build or update any containers once I install RC3 or RC4.  If I switch back to the standard build, no issues.  I even tried switching the older K2200 card out with a 1050 Ti.   

 

I tried multiple different containers, but all spit out errors.   Any insights? 

 


root@localhost:# /usr/local/emhttp/plugins/dynamix.docker.manager/scripts/docker run -d --name='binhex-plexpass' --net='host' --log-opt max-size='50m' --log-opt max-file='1' -e TZ="America/New_York" -e HOST_OS="Unraid" -e 'TRANS_DIR'='/config/transcode' -e 'UMASK'='000' -e 'PUID'='99' -e 'PGID'='100' -e 'NVIDIA_DRIVER_CAPABILITIES'='all' -e 'NVIDIA_VISIBLE_DEVICES'='GPU-c8814be1-3d0f-15fa-c1c4-bf0f4aa47348' -v '/mnt/user':'/media':'rw' -v '/mnt/cache/appdata/binhex-plexpass':'/config':'rw' --runtime=nvidia 'binhex/arch-plexpass' 
panic: runtime error: invalid memory address or nil pointer dereference
[signal SIGSEGV: segmentation violation code=0x1 addr=0xb2 pc=0x8b00ec]

goroutine 1 [running]:
github.com/docker/cli/vendor/github.com/spf13/pflag.(*FlagSet).VarPF(0xc0000c6800, 0x2197420, 0xc00021d1c8, 0x1e4316a, 0x6, 0x0, 0x0, 0x1e71894, 0x1f, 0x1e7622f)
/go/src/github.com/docker/cli/vendor/github.com/spf13/pflag/flag.go:817 +0x6c
github.com/docker/cli/vendor/github.com/spf13/pflag.(*FlagSet).VarP(...)
/go/src/github.com/docker/cli/vendor/github.com/spf13/pflag/flag.go:825
github.com/docker/cli/vendor/github.com/spf13/pflag.(*FlagSet).StringVar(0xc0000c6800, 0xc00021d1c8, 0x1e4316a, 0x6, 0xc0000f2da0, 0xd, 0x1e71894, 0x1f)
/go/src/github.com/docker/cli/vendor/github.com/spf13/pflag/string.go:37 +0x9d
github.com/docker/cli/cli.setupCommonRootCommand(0xc0003b4f00, 0x1c448a0, 0xbd, 0xbe)
/go/src/github.com/docker/cli/cli/cobra.go:24 +0xe1
github.com/docker/cli/cli.SetupRootCommand(0xc0003b4f00, 0xc00002ebd0, 0xc0001b1df0, 0x2)
/go/src/github.com/docker/cli/cli/cobra.go:55 +0x2f
main.newDockerCommand(0xc0001540f0, 0x11)
/go/src/github.com/docker/cli/cmd/docker/docker.go:54 +0x233
main.runDocker(0xc0001540f0, 0x2150ba0, 0xc00000e020)
/go/src/github.com/docker/cli/cmd/docker/docker.go:250 +0x40
main.main()
/go/src/github.com/docker/cli/cmd/docker/docker.go:292 +0xf1

The command failed.

 

So as an update, I gave it one last shot and before I started up any containers I ran the Docker Safe Permissions tool. Everything works as planned now. 

 

Appreciate all of the work done. 

Link to comment

rc5 doesn't work for me. several error messages about not finding unraid modules

updating module dependencies...

file not found

(sry, don't have a screencapture, getting server back to work was more important at the moment)

when booted i don't have any network and emhttpd crashes with a segfault

 

back to rc4 works fine

Link to comment
11 hours ago, Benni-chan said:

rc5 doesn't work for me. several error messages about not finding unraid modules

updating module dependencies...

file not found

(sry, don't have a screencapture, getting server back to work was more important at the moment)

when booted i don't have any network and emhttpd crashes with a segfault

 

back to rc4 works fine

Tested / updated to rc5 myself yesterday and it worked fine. Sounds like the plugin didn't finish downloading the files. Please try it again and check the plugin shows all the files completed.

Link to comment
4 minutes ago, ashman70 said:

I have a question, if I have an Nvidia P2000, do I still need to add the Nvdia plugin to my Plex server to use it, or is it natively supported in Plex under unRAID?

You need the plugin, and use it to install an Nvidia variant of Unraid - that is exactly what this plugin is for.

Link to comment

Any reason why my 980 is not seen anymore? 

Version: 6.8.0-rc5

 

Nvidia Driver Version:430.50

GPU 0 Model & Bus:No devices were found

 

IOMMU group 32:[10de:13c0] 43:00.0 VGA compatible controller: NVIDIA Corporation GM204 [GeForce GTX 980] (rev a1)

[10de:0fbb] 43:00.1 Audio device: NVIDIA Corporation GM204 High Definition Audio Controller (rev a1)

skynet-ur-diagnostics-20191105-0928.zip

Edited by turnipisum
Link to comment

First, thank you for making this. It has been a huge help offloading my CPU with GPU in Plex.

 

However, I am having an issue. I have a Geforce GTX 1070 and a Quadro P2000. I use the Quadro for pass through to the Plex docker and I want to use the 1070 for pass through to my VMs. Here is the issue, When the 1070 is installed in my machine, Plex uses it instead of the Quadro. "watch nvidia-smi" confirms that transcoding is happening on the 1070. I have tried switching PCI ports and have verified that the GPU UUID in the Plex container is the one for the Quadro. 

 

Any help would be greatly appreciated.

Edited by veritas2884
spelling
Link to comment
1 hour ago, veritas2884 said:

First, thank you for making this. It has been a huge help offloading my CPU with GPU in Plex.

 

However, I am having an issue. I have a Geforce GTX 1070 and a Quadro P2000. I use the Quadro for pass through to the Plex docker and I want to use the 1070 for pass through to my VMs. Here is the issue, When the 1070 is installed in my machine, Plex uses it instead of the Quadro. "watch nvidia-smi" confirms that transcoding is happening on the 1070. I have tried switching PCI ports and have verified that the GPU UUID in the Plex container is the one for the Quadro. 

 

Any help would be greatly appreciated.

 

I think you can change the "all" to the # of the card in the NVIDIA_DRIVER_CAPABILITIES variable of your PLEX docker.  That will isolate what NVIDIA card is passthru to this card instead of passing it all.

Link to comment
2 hours ago, Pducharme said:

 

I think you can change the "all" to the # of the card in the NVIDIA_DRIVER_CAPABILITIES variable of your PLEX docker.  That will isolate what NVIDIA card is passthru to this card instead of passing it all.

Thank you for the reply. I tried to change the value to 1 from all, and got a startup error from plex. Here are my settings.

plex settings.JPG

unraid nvidia.JPG

watch nvidia.JPG

Link to comment
2 minutes ago, veritas2884 said:

Thank you for the reply. I tried to change the value to 1 from all, and got a startup error from plex. Here are my settings.

plex settings.JPG

unraid nvidia.JPG

watch nvidia.JPG

I have a similar setup. I never encountered this issue, but I’ve always had the gaming GPU stubbed and passed to the VM. If you stub the card, the NVIDIA plugin won’t see it at all, thus eliminating any possibility of it using the wrong card. I suggest trying that if you don’t need the 1070 outside VM usage.

Link to comment
14 minutes ago, JasonM said:

I have a similar setup. I never encountered this issue, but I’ve always had the gaming GPU stubbed and passed to the VM. If you stub the card, the NVIDIA plugin won’t see it at all, thus eliminating any possibility of it using the wrong card. I suggest trying that if you don’t need the 1070 outside VM usage.

Thank you! No, I won't need that outside of VM. How do you stub the card? Also will it stay locked even if the VM is off?

 

EDIT: I looked up how to stub pci devices. I added pci-stub.ids=10de:10f0 to the flash drive append area. I will give this a go and see what happens. Thanks again.

Edited by veritas2884
Link to comment
16 minutes ago, dudarus1 said:

I have exactly the same issue as above running a gtx 1060 and a quaddro p400. doesnt matter what UUID i put into the container it always picks the gtx 1060. thought i was doing something wrong but ive been trying for days to get it to transcode on the p400

So the stubbing that was mentioned didn't seem to work, but I am not sure that I did it properly. However, I will say that if you stop the Plex container, start a VM that uses the 1060, then start the Plex container again it will only use the Quadro. Downfall here is that you need to leave the VM running to all the time or it will fall back to the 1060.

Edited by veritas2884
Link to comment
  • trurl locked this topic
Guest
This topic is now closed to further replies.