[Plugin] Nvidia-Driver


ich777

Recommended Posts

8 hours ago, SamIAm said:

is there any way to update the nvidia driver through a user script?

Not really.

 

8 hours ago, SamIAm said:

I would do it through your plugin on reboot but my router is a pfSense vm which as you can imagine makes it a little difficult to be connected to the internet while trying to update.

I've already heard that people running pfSense as a VM on Unraid and thought of that, I will see and update the plugin in the next few days so that you have a Download in the plugin to download it before the restart.

 

For now, but that's really not an elegant solution, but you can try to delete the plugin reboot reinstall the plugin and then restart again.

 

EDIT: But one problem I won't be able to solve with this, if there is an update of Unraid you have to always reinstall the Plugin since it can't get the new version on boot.

 

EDIT2: Wouldn't such a device solve this "issue": Click thanks go out to @binhex who pointed me to this. :)

  • Like 1
  • Thanks 1
Link to comment
6 hours ago, ich777 said:

Not really.

 

I've already heard that people running pfSense as a VM on Unraid and thought of that, I will see and update the plugin in the next few days so that you have a Download in the plugin to download it before the restart.

 

For now, but that's really not an elegant solution, but you can try to delete the plugin reboot reinstall the plugin and then restart again.

 

EDIT: But one problem I won't be able to solve with this, if there is an update of Unraid you have to always reinstall the Plugin since it can't get the new version on boot.

 

EDIT2: Wouldn't such a device solve this "issue": Click thanks go out to @binhex who pointed me to this. :)

 

Thabks for looking into this @ich777! I like the idea of the download within the plugin. 
 

I encountered the need to uninstall and reinstall your plugin when upgrading from 6.9 rc2 to 6.9, wasn’t too much of an inconvenience for me once I figured it out. 
 

I also plan on going to full metal pfSense soon, this wasn’t the only thing that becomes a minor annoyance when running it as a vm. 

Edited by SamIAm
Grammar
  • Like 1
Link to comment
On 3/4/2021 at 1:51 PM, Glasti said:

The issue/error has returned again. 

Testing with not transcoding to ram but to 1 of my cache pools. If that doesnt help i am going to run a MemTest. 


 

Update: Hopefully the last one. 
I haven't run Memtest yet. But tried to go back to RC2 but that didnt solve the issue.. 
So i decided to uninstall the plugin and upgrade to Stable again. After updating OS i grabbed the plugin and i saw that newer drivers where added, i was using 455.** before. 

Now i am running driver 460.56 and the issue hasnt occured for 28 hours, which is great.. 

Of course i wasnt smart enough to grab the driver i was using before after reinstalling the plugin to see if it was an issue with upgrading from RC2 to Stable without reinstalling the plugin and/or driver.

But i am happy it (hopefully) is solved, and if hopefully this can help anyone else that sees this issue.
 

Edited by Glasti
  • Like 1
Link to comment
28 minutes ago, thymon said:

nothing works on my EMBY :( GPU not detected on EMBY

"Binhex-Emby"

 

I don't know why... 

Can you try the emby/embyserver build from the official source just for troubleshooting reasons (you find it also in the CA App)?

 

I have no problem with the above mentioned Container:

grafik.png.14e59e47b06395c69c991bfaf756d014.png

Link to comment

Hi, I'm getting the "No devices were found" error in the plugin and when trying "nvidia-smi" even though it detects my Quadro K4000 in system devices and with "lscpi" in terminal. I'm on stable Unraid 6.9.0 and Nvidia-driver version 2021.03.07. I am using a Gigabyte Z77X-UP4 TH motherboard and have Vt-x and Vt-D enabled in the bios. I tried it with and without the card plugged into a display. I've also tried bios resets, many restarts, re-seating the card, legacy boot mode, checking and unchecking vfio bind as advised by previous posts. I've also confirmed that the card can be detected in a Windows VM. Any advice would be greatly appreciated. Thanks!

 

Edit: I'm also getting a weird error in the logs RmInitAdapter failed and failed to copy vbios to system memory whenever i tried to run nvidia-smi or open the nvidia-driver settings page, as per the image.

Also the card outputs properly to the external monitor and functions normally in a windows computer.

 

systemDevices.PNG

nvidiaPluginNoDevices.PNG

weirdLog.PNG

Edited by quizzy99
Included log
Link to comment
On 3/5/2021 at 7:28 PM, SiRMarlon said:

Just as and FYI on this system.

 

AMD Ryzen 7 2700x / Asrock B450 ITX / Nvidia Quadro P2000 its been about 24 hours since I installed the card back into the system. I disabled C-States in the BIOS, I am running the card with the SMI instance as mentioned above and the card has not fallen off the bus like it was doing before. I am going to let it run through the weekend and if by Monday it has not fallen off the bus I will go ahead and re-configure PLEX to use the card and put it under load with a transcode to see if still stays stable. 

 

I am on Driver version 460.56

 

It's been 36hrs and the GPU has been stable and has not fallen off the bus like it was doing before! I've gone ahead and configured Plex hardware transcoding and I have 4 streams going right now. All of them are trancoding 4K HDR to 1080p SDR

 

image.png.b5ce4b5086365f5420abee6e86da9f02.png

 

image.png.0d2ce9c151ada7093b03dfb351b281da.png

 

image.png.166146841e88d866f35edd362c903355.png

 

image.png.fe31cc5af48caf49ec426ccee3cca451.png

 

image.png.abff76b0209dde894880c94bb15c8622.png

 

As you can see HW transcoding is working. So I have the movies on repeat and I am going to let this run all night hopefully when I wake up the server is still running and has not blown up! 😂

 

image.png.23a3596ed7643ebb658da7f41b9ce117.png

 

So for those of you who happen to be on AMD Chips and are having issues go into your BIOS and disabled C-States as this seems to be working. But we'll see where I am at tomorrow morning! 

 

Also I do have my Plex transcoding quality set to "MAKE MY CPU HURT"  the server never sees this much workload so this is def making the system work.

 

image.png.af3cd3a208d3d4abd6562b5a68719cbb.png

 

 

Edited by SiRMarlon
  • Like 1
Link to comment
7 hours ago, quizzy99 said:

Hi, I'm getting the "No devices were found" error in the plugin and when trying "nvidia-smi" even though it detects my Quadro K4000 in system devices and with "lscpi" in terminal. I'm on stable Unraid 6.9.0 and Nvidia-driver version 2021.03.07. I am using a Gigabyte Z77X-UP4 TH motherboard and have Vt-x and Vt-D enabled in the bios. I tried it with and without the card plugged into a display. I've also tried bios resets, many restarts, re-seating the card, legacy boot mode, checking and unchecking vfio bind as advised by previous posts. I've also confirmed that the card can be detected in a Windows VM. Any advice would be greatly appreciated. Thanks!

 

Edit: I'm also getting a weird error in the logs RmInitAdapter failed and failed to copy vbios to system memory whenever i tried to run nvidia-smi or open the nvidia-driver settings page, as per the image.

Also the card outputs properly to the external monitor and functions normally in a windows computer.

 

systemDevices.PNG

nvidiaPluginNoDevices.PNG

weirdLog.PNG

I had a similar issues fixed it by doing the following. 

 

- delete the nvidia plugin

- grab fresh unraid kernel

- install plugin and latest driver. 

 

Maybe it helps. 

 

I have not changed any hardware  settings. 

 

Ryzen 3700x, strix b450-f and a gtx 1660

  • Thanks 1
Link to comment
5 minutes ago, winterkid310 said:

hi all.
i am having  where by the nvidia is not detecting installed GPU. 

In your syslog I see a lot of this messages:

 

Quote

Mar  8 16:41:51 Tower kernel: NVRM: GPU 0000:82:00.0: GPU does not have the necessary power cables connected.
Mar  8 16:41:51 Tower kernel: NVRM: GPU 0000:82:00.0: RmInitAdapter failed! (0x25:0x1c:1262)
Mar  8 16:41:51 Tower kernel: NVRM: GPU 0000:82:00.0: rm_init_adapter failed, device minor number 0

 

 

Also is this done by accident in your syslinux.cfg:

BOOT_IMAGE=/bzimage vfio-pci.ids=8086:10bc initrd=/bzroot

 

I can't even find a device in your system that has the id 8086:10bc

Link to comment

Appreciate all the support in this thread.

 

I can't seem to find an answer to my question however and clearly don't understand containers and GPUs enough to work this out myself.

 

When sharing a GPU between multiple containers (E.G. Folding@home, Plex) is it possible to limit GPU resources per container? If I only wanted to offer 10% of my GPU's resources to folding and have the rest available to other containers - is this possible?

Link to comment
47 minutes ago, alpha302 said:

When sharing a GPU between multiple containers (E.G. Folding@home, Plex) is it possible to limit GPU resources per container?

As far as I know this isn't possible, you can assign a card to multiple containers, limit the capabilities to a container but not limit the usage of it per container.

You also can assign certain CPU cores to a container and share also the same core in multiple containers but not limit the usage. See it like this: GPU = 1 Core...

Hope this makes somehow sense to you, this is a highly simplified answer.

  • Like 1
Link to comment
10 hours ago, SiRMarlon said:

 

It's been 36hrs and the GPU has been stable and has not fallen off the bus like it was doing before! I've gone ahead and configured Plex hardware transcoding and I have 4 streams going right now. All of them are trancoding 4K HDR to 1080p SDR

 

image.png.b5ce4b5086365f5420abee6e86da9f02.png

 

image.png.0d2ce9c151ada7093b03dfb351b281da.png

 

image.png.166146841e88d866f35edd362c903355.png

 

image.png.fe31cc5af48caf49ec426ccee3cca451.png

 

image.png.abff76b0209dde894880c94bb15c8622.png

 

As you can see HW transcoding is working. So I have the movies on repeat and I am going to let this run all night hopefully when I wake up the server is still running and has not blown up! 😂

 

image.png.23a3596ed7643ebb658da7f41b9ce117.png

 

So for those of you who happen to be on AMD Chips and are having issues go into your BIOS and disabled C-States as this seems to be working. But we'll see where I am at tomorrow morning! 

 

Also I do have my Plex transcoding quality set to "MAKE MY CPU HURT"  the server never sees this much workload so this is def making the system work.

 

image.png.af3cd3a208d3d4abd6562b5a68719cbb.png

 

 

 

 

All systems go! No issues to report, the server handled it like a champ, looks like we are good! Thanks for your help! 

Edited by SiRMarlon
  • Like 1
Link to comment
2 hours ago, ich777 said:

As far as I know this isn't possible, you can assign a card to multiple containers, limit the capabilities to a container but not limit the usage of it per container.

You also can assign certain CPU cores to a container and share also the same core in multiple containers but not limit the usage. See it like this: GPU = 1 Core...

Hope this makes somehow sense to you, this is a highly simplified answer.

 

Yes, that makes sense. Thanks for the prompt response!

  • Like 1
Link to comment

So I was on 6.9 beta 35 with the Plugin and all worked fine, booted to GUI locally on the server without issue, while still being used in plex and tidarr.  Now I updated to 6.9.0 Stable and upon reboot it went into the GUI as expected, then I re installed the plugin, rebooted and I get the normal output for loading unraid but when it comes time to show the GUI i get a blank screen with a blinking cursor in the upper left corner, but everything else works, its just no local GUI. I get GUI from my web browser, plex transcodes as before, tidarr encodes as before, VM pass though still works as expected. Everything works as before just the Local GUI no longer shows. I boot to Legacy mode, have 2 card installed, PCI slot 1 is the card used for Unraid/Plex/Tidarr. PCI slot 4 is passed through to my gamming vm.

Link to comment

Hi ich777,    Thank you for taking the time to review my issue.  I did force the container update as suggested on the first page of the thread.  It seems after switching from linxserver container to binhex-plexpass, the transcoder was having permission issues.  Ultimately, I deleted the transcode variable from the container and replaced it with a path variable pointing to /tmp.   Hardware accelerated encoding is now working.   Thanks again for your great support and hard work on the plug-in!

Link to comment
17 hours ago, Glasti said:

I had a similar issues fixed it by doing the following. 

 

- delete the nvidia plugin

- grab fresh unraid kernel

- install plugin and latest driver. 

 

Maybe it helps. 

 

I have not changed any hardware  settings. 

 

Ryzen 3700x, strix b450-f and a gtx 1660

 

I appreciate the tip. I just tried booting with a fresh copy of Unraid 6.9 straight from the website and installing the plugin, but unfortunately I got the same "No devices were found" error.

Link to comment

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Restore formatting

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.