[Plugin] Linuxserver.io - Unraid Nvidia


Recommended Posts

5 minutes ago, jebusfreek666 said:

Are you able to view active processes in nvidia-smi?

 

Would you mind sharing your template? Maybe I messed it up somewhere. I kind of doubt that is the issue though, as I just copy paste my settings to the linuxserver docker and it is working. What version of unraid are you running? And what GPU?

Yes i can see everything transcoding both reported in plex as well as Nvidia-smi 

 

I am running 6.8.1 with a P2000

 

You should show us a screenshot of your Nvidia screen as well as settings for docker. Also just to make sure hardware accel is on in plex itself.

Link to comment

I will screenshot when I get home. At work now. I am on 6.8, but I don't think that is the issue... I am also running a p2000. Like I said, I am fairly sure things are setup correctly as I ran through them 3 times and copy/paste the settings to the linuxserver docker and it worked. But I will screenshot in the AM.

Link to comment
8 hours ago, Jhp612 said:

Yes i can see everything transcoding both reported in plex as well as Nvidia-smi 

 

I am running 6.8.1 with a P2000

 

You should show us a screenshot of your Nvidia screen as well as settings for docker. Also just to make sure hardware accel is on in plex itself.

 

Here is my setup. Do you guys see any glaring mistakes? Can't figure out why I can't get this to work with binhex plex (not binhex plex pass version). I have plex pass, hw transcoding is on, I think my settings are correct, and with the same settings it works in the linuxserver container?

 

 

Screenshot 2020-01-31 at 7.03.37 AM.png

Screenshot 2020-01-31 at 6.59.16 AM.png

Screenshot 2020-01-31 at 6.59.09 AM.png

Link to comment

For those who may be experiencing the same issue with the binhex container, I have found a solution. For some dumb reason that I can't explain, you have to enter both variable twice. I added them both twice and now it works. Thinking maybe I messed up the first time, I deleted the first set of variables, leaving only one set. After doing that, the hardware transcoding stopped working again. Then I added back the second identical set of variables, and it works again. I can't explain it, but there it is....

 

 

Screenshot 2020-01-31 at 8.25.51 AM.png

Link to comment
2 hours ago, jebusfreek666 said:

For those who may be experiencing the same issue with the binhex container, I have found a solution. For some dumb reason that I can't explain, you have to enter both variable twice. I added them both twice and now it works. Thinking maybe I messed up the first time, I deleted the first set of variables, leaving only one set. After doing that, the hardware transcoding stopped working again. Then I added back the second identical set of variables, and it works again. I can't explain it, but there it is....

 

 

Screenshot 2020-01-31 at 8.25.51 AM.png

 

I can pretty much guarantee that is the wrong conclusion. My guess is that you have an extra space somewhere you can't see. it would help if you posted your docket run command

  • Thanks 1
Link to comment
On 1/26/2020 at 8:32 PM, phasetwo said:

 

Sorry for bringing this up again but this seems to still be an issue.

I'm running 6.8 rc7 with the current binhex plex docker and my P600 does HW transcoding for everything except HEVC 10 bit HDR.  

I'm aware that HEVC 10bit 4:4:4 NVDEC is not supported. However, a file with 4:2:0 chroma subsampling should be decoded by the card but is not. Does anyone have a clue as to why that is?

 

SOLVED: Since updating to 6.8.1 to the P600 transcodes large HEVC 10bit HDR files. I suspect that the driver update from 440.36 to 440.44 might have solved the issue.

  • Like 1
Link to comment

Hi!

 

I upgraded unraid from 6.7-something i think to 6.8.1 and then i downloaded and installed unraid nvidia for 6.8.1 and rebooted. After that I can't login to the webGUI. If I enter root credentials in the new fancy webGUI loginscreen it just seems to reload, if I enter some other account I get "Invalid Username or Password" error.

SSH with root or the other account works fine though.

 

Any one got a clue what to do? Can I revert back to vanilla unraid from the terminal somehow and see if I can login after that?

Link to comment
11 minutes ago, CDivander said:

Any one got a clue what to do? Can I revert back to vanilla unraid from the terminal somehow and see if I can login after that?

To revert, download the 6.8.1 zip file from limetech's download page and overwrite all of the bz* files on the flash drive with the ones in the zip and then reboot.

Link to comment

Hi all - is anyone using this with Shinobi? I turned on hardware acceleration within Shinobi and selected cuvid (NVIDIA NVENC) under Acceleration Engine. There is a "HWAccel Device" which wants a path and a "Use coProcessor" that I also left blank.

Link to comment

Possibly a bit of a noob question - I'm currently nice and stable on 6.7.2 with driver 430.16 on a Quadro P620.

 

I'm wanting to roll out the update to 6.8.2 but noticed the latest nvidia driver through this plugin is 6.8.1 - is there any issue in updating and using that driver or should I wait until the driver is updated for 6.8.2 as well?

Link to comment

You can't update unRaid versions until such point as lsio releases an update to the plugin.  (Don't use this plugin, so IDK if this has already happened or not yet).   IE: Once you're using unRaid nVidia or unRaid DVB, then you will remain using them and never upgrade the OS through "normal" means

Link to comment
9 minutes ago, Squid said:

You can't update unRaid versions until such point as lsio releases an update to the plugin.  (Don't use this plugin, so IDK if this has already happened or not yet).   IE: Once you're using unRaid nVidia or unRaid DVB, then you will remain using them and never upgrade the OS through "normal" means

Thanks squid - so what's the general OS update procedure in that case?

 

  • Revert to stock build, update, lsio driver?
  • Pre-update via lsio to new driver then update OS?
  • OS as normal then patch driver in?
  • Other?

 

Apologies if this is obvious but I've not seen anything about correct update procedure and don't fancy bricking my server when I've not really got the time to sort it out if it bricks

Link to comment
1 minute ago, Squid said:

Wait for the plugin to be updated then update via the plugin.  Nothing else needs to be done

And this will update the OS as well? I thought the plugin was just for the nvidia driver?

 

So in this case, I could update to 6.8.1 via the plugin now? But just can't update to the stock 6.8.2 build via the builtin updater

Link to comment
2 minutes ago, headbocks said:

And this will update the OS as well? I thought the plugin was just for the nvidia driver?

 

The driver is intimately tied to the OS.

3 minutes ago, headbocks said:

So in this case, I could update to 6.8.1 via the plugin now?

Yes

3 minutes ago, headbocks said:

But just can't update to the stock 6.8.2 build via the builtin updater

No

  • Like 1
Link to comment
4 minutes ago, headbocks said:

And this will update the OS as well? I thought the plugin was just for the nvidia driver?

 

So in this case, I could update to 6.8.1 via the plugin now? But just can't update to the stock 6.8.2 build via the builtin updater

The "plugin" is not the driver, it is a custom build of the OS that includes the driver. I think this is explained pretty well in the beginning of this thread.

Link to comment

 

43 minutes ago, headbocks said:

So in this case, I could update to 6.8.1 via the plugin now?

Be aware that 6.8.1 has an issue with the intel drivers that *can* lock your system on boot (not every boot and it is the same with stock Unraid).  Mine had that issue, so I'm on 6.8.0.

Link to comment

v6.8.2 uploaded.

 

Delayed for a few reasons, had problems (and still do) with the nvidia container runtime, worked around it in the end, but not a long term solution looking forward, I'm working like a dog at the moment as my current real life job finishes in 2 days and I'm having to put a ton of extra hours in, wife a bit ungainly at the moment as very heavily pregnant so I'm having to do a bit more for our existing beast, and to add to that bass_rock has been away for work, so kind of a perfect storm of not having much time to sit down with this, although I have been trying to get it working every chance I've had.

 

Anyways, I've tested this version, think everything is working, and I believe all the out of tree drivers are squared away.  Last version (v6.8.1) might have been missing the Intel 1gb driver as I hadn't realised that it was different to the 10gb driver.

  • Like 6
  • Thanks 10
  • Haha 1
Link to comment
7 minutes ago, CHBMB said:

v6.8.2 uploaded.

 

Delayed for a few reasons, had problems (and still do) with the nvidia container runtime, worked around it in the end, but not a long term solution looking forward, I'm working like a dog at the moment as my current real life job finishes in 2 days and I'm having to put a ton of extra hours in, wife a bit ungainly at the moment as very heavily pregnant so I'm having to do a bit more for our existing beast, and to add to that bass_rock has been away for work, so kind of a perfect storm of not having much time to sit down with this, although I have been trying to get it working every chance I've had.

 

Anyways, I've tested this version, think everything is working, and I believe all the out of tree drivers are squared away.  Last version (v6.8.1) might have been missing the Intel 1gb driver as I hadn't realised that it was different to the 10gb driver.

Great Job Sir!!!! Thank you for your continued efforts!!!!

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