[Plugin] Linuxserver.io - Unraid Nvidia


2468 posts in this topic Last Reply

Recommended Posts

  • Replies 2.5k
  • Created
  • Last Reply

Top Posters In This Topic

Top Posters In This Topic

Popular Posts

DEPRECATED

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 wo

DEPRECATED

Posted Images

4 hours ago, d2dyno said:

Not using jumbo frames is a workaround, not a fix. I use that MTU on purpose, and never had issues except with this plugin.

It could also be the bonding that is causing an issue here.

We've never been able to replicate these issues which is why these are the only way's we know to fix it. But it seems like we haven't been able to replicate these issues because no one within the team that uses the nvidia plugin has any 10gbit networking kit or uses bonding (to my knowledge).

Link to post

In case it helps to rule anything out, I have a bond between an onboard 1Gb NIC and one port in a Dual Port 10Gb Nic (running also at 1Gb).  No issues with downloading the plugin.

Edited by Marshalleq
Link to post

I am trying to get this plugin to work.  I am trying to use a Nvidia GeForce GTX 1050 card and using the Nvidia 6.8.3 version however when i install and reboot i am getting the following error:

 

NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver.  Make sure that the latest NVIDIA driver is installed and running.

The driver version listed is 440.59 and i am not sure how i would update this if it is not the latest driver?

Link to post

Since updating I see this over and over until it fills my log:

 

Mar 13 16:28:03 NAS kernel: caller _nv000908rm+0x1bf/0x1f0 [nvidia] mapping multiple BARs
Mar 13 16:28:06 NAS kernel: resource sanity check: requesting [mem 0x000c0000-0x000fffff], which spans more than PCI Bus 0000:00 [mem 0x000d0000-0x000dffff window]

Anyone else seeing this? I'm just using an old 750 ti.

Link to post

I was having trouble getting the new driver installed as well. I do have a 10G network card installed and so I went down that path trying to figure out if it was the issue. I did the same sort of troubleshooting as others. Changed from static to DHCP, messed with MTU and verified I wasn't doing any bonding. After all that I remembered I had re-installed my Sophos Endpoint protection a few days ago on my main PC I use to access the GUI. I disabled my endpoint protection it and it downloaded and installed the driver without issue. At no time did it ever tell me it was blocking anything. Just wanted to throw this out there as another possible work around for people having issues. Im sure this is probably mentioned previously somewhere in this thread, and if so I must have missed it. I would not have thought my endpoint protection on my PC would affect the GUI of Unraid in my browser when it tried to download and install a driver. You learn something new everyday. 

Link to post

Currently facing an issue where hardware transcoding isn't working after a random amount of time. On my samsung tv, the logs were showing that the transcoder was dead. Streaming from a pc, I don't see anything relevant in the logs. Thought it might be due to the transcoding directory being full, but checked and changed a couple ways and that doesn't seem to be related.

Seemed to occur out of the blue on 6.8.2 with no relevant changes on my end, so I upgraded to 6.8.3 but the issue still occurs.
Attaching log file from session when streaming to pc where the video just halts (like it was paused, no errors on screen)

 

GPU is a GTX 1650. This is out output of "watch nvidia-smi" even when the video is halted:
 

+-----------------------------------------------------------------------------+
| NVIDIA-SMI 440.59       Driver Version: 440.59       CUDA Version: 10.2     |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  GeForce GTX 1650    Off  | 00000000:0C:00.0 Off |                  N/A |
| 40%   35C    P0    19W /  75W |     76MiB /  3903MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID   Type   Process name                             Usage      |
|=============================================================================|
|    0      3355      C   /usr/lib/plexmediaserver/Plex Transcoder      64MiB |
+-----------------------------------------------------------------------------+

 

When streaming from the TV, the nvidia-smi doesn't show anything after the video stops / error is shows on the tv.

 

Any help would be appreciated.

Plex Media Server Logs_2020-03-16_12-40-26.zip

Edited by tlrmcknz
surround in code block
Link to post
On 3/15/2020 at 3:15 PM, Tragik said:

I was having trouble getting the new driver installed as well. I do have a 10G network card installed and so I went down that path trying to figure out if it was the issue. I did the same sort of troubleshooting as others. Changed from static to DHCP, messed with MTU and verified I wasn't doing any bonding. After all that I remembered I had re-installed my Sophos Endpoint protection a few days ago on my main PC I use to access the GUI. I disabled my endpoint protection it and it downloaded and installed the driver without issue. At no time did it ever tell me it was blocking anything. Just wanted to throw this out there as another possible work around for people having issues. Im sure this is probably mentioned previously somewhere in this thread, and if so I must have missed it. I would not have thought my endpoint protection on my PC would affect the GUI of Unraid in my browser when it tried to download and install a driver. You learn something new everyday. 

I tried disabling the endpoint protection i had on the pc i was using to access the GUI but still no luck.

Link to post

Need a little help. I updated to the linuxserverio Unraid 6.8.3. Main logs and console complain that the nvidia driver couldn't connect to my 1660TI and it thought that another driver was grabbing it. I do have virtio grabbing the card for when I have a Win10 vm booted. Is there anyway to get this to share between them or get virtio to talk to the nvidia driver in a proxy sort of way?

Link to post
6 hours ago, ghstridr said:

Need a little help. I updated to the linuxserverio Unraid 6.8.3. Main logs and console complain that the nvidia driver couldn't connect to my 1660TI and it thought that another driver was grabbing it. I do have virtio grabbing the card for when I have a Win10 vm booted. Is there anyway to get this to share between them or get virtio to talk to the nvidia driver in a proxy sort of way?

The first post explains it. You can't use the same gpu with docker and vm at the same time. VMs require exclusive access

Link to post
On 3/13/2020 at 4:32 PM, O_M_R said:

Since updating I see this over and over until it fills my log:

 


Mar 13 16:28:03 NAS kernel: caller _nv000908rm+0x1bf/0x1f0 [nvidia] mapping multiple BARs
Mar 13 16:28:06 NAS kernel: resource sanity check: requesting [mem 0x000c0000-0x000fffff], which spans more than PCI Bus 0000:00 [mem 0x000d0000-0x000dffff window]

Anyone else seeing this? I'm just using an old 750 ti.

 

I see this too ... it's infrequent unless I also install the GPU Statistics Plugin, then it's constant.  GTX 1060 here.

 

With the the GPU Statistics Plugin:

Mar 19 20:32:46 Tower kernel: caller _nv000908rm+0x1bf/0x1f0 [nvidia] mapping multiple BARs
Mar 19 20:32:49 Tower kernel: resource sanity check: requesting [mem 0x000c0000-0x000fffff], which spans more than PCI Bus 0000:00 [mem 0x000c0000-0x000dffff window]

A few minutes later after uninstalling the GPU Statistics Plugin (no caller line in the log):

Mar 19 20:37:08 Tower kernel: resource sanity check: requesting [mem 0x000c0000-0x000fffff], which spans more than PCI Bus 0000:00 [mem 0x000c0000-0x000dffff window]
M

 

Link to post
On 3/19/2020 at 8:45 PM, JesterEE said:

 

I see this too ... it's infrequent unless I also install the GPU Statistics Plugin, then it's constant.  GTX 1060 here.

 

With the the GPU Statistics Plugin:


Mar 19 20:32:46 Tower kernel: caller _nv000908rm+0x1bf/0x1f0 [nvidia] mapping multiple BARs
Mar 19 20:32:49 Tower kernel: resource sanity check: requesting [mem 0x000c0000-0x000fffff], which spans more than PCI Bus 0000:00 [mem 0x000c0000-0x000dffff window]

A few minutes later after uninstalling the GPU Statistics Plugin (no caller line in the log):


Mar 19 20:37:08 Tower kernel: resource sanity check: requesting [mem 0x000c0000-0x000fffff], which spans more than PCI Bus 0000:00 [mem 0x000c0000-0x000dffff window]
M

 

Just wanted to drop you a quick thank you, I'll turn off the GPU statistics Plugin for now. I was wondering if that was it, I just haven't had time to look at it yet.

Link to post
5 hours ago, O_M_R said:

Just wanted to drop you a quick thank you, I'll turn off the GPU statistics Plugin for now. I was wondering if that was it, I just haven't had time to look at it yet.

No problem.

 

I just want to be clear, this seems to be an underlying problem in the build ... the a GPU Stats plug-in just makes it more apparent.

 

-JesterEE

Link to post

Just curious, has anyone confirmed this plugin works with a 1660 Super? From the current driver version (440.59), it would seem that would be the case. If someone has first hand experience though with this card, it would be great before placing my order for it. Thanks!

Link to post

I am at a loss. I have been unsuccessful at getting my GTX1070 to work on my Unraid 6.8.3 server. Want to use it on Handbrake and then Plex, no VMs. I have been at it for a couple of months and my build is defaulting to CPU transcoding, not GPU. Here's what I have tried.

1. Latest unraid and nvidia plug-in version. It sees it as GPU0 and gives a UUID number

2. Using Djaydevs version of Handbrake to support Nvidia GPU

3. Set Handbrakes NVIDIA_DRIVER_CAPABILITIES = all

4.Set the extra parm: --runtime=nvidia

5. Verified there is no trailing space after the GPU UUID

6. Confirmed in my BIOS that the server graphics are using the on-board  GPU and not Nvidia

7. Confirmed ASRock supports ths GPU

 

...what am I missing here?

Unraid: 6.8.3

System board: ASRock x570 Phantom Gaming 4, BiOS P1.30

CPU [dont laugh]: AMD Ryzen 3 3200G with Radeon Vega Grahics @3600

GPU: Nvidia GTX 1070 w/ driver version 440.59

Mem: 8GiB DDR4

 

 

Link to post
1 hour ago, endessunraid said:

I am at a loss. I have been unsuccessful at getting my GTX1070 to work on my Unraid 6.8.3 server. Want to use it on Handbrake and then Plex, no VMs. I have been at it for a couple of months and my build is defaulting to CPU transcoding, not GPU. Here's what I have tried.

1. Latest unraid and nvidia plug-in version. It sees it as GPU0 and gives a UUID number

2. Using Djaydevs version of Handbrake to support Nvidia GPU

3. Set Handbrakes NVIDIA_DRIVER_CAPABILITIES = all

4.Set the extra parm: --runtime=nvidia

5. Verified there is no trailing space after the GPU UUID

6. Confirmed in my BIOS that the server graphics are using the on-board  GPU and not Nvidia

7. Confirmed ASRock supports ths GPU

 

...what am I missing here?

Unraid: 6.8.3

System board: ASRock x570 Phantom Gaming 4, BiOS P1.30

CPU [dont laugh]: AMD Ryzen 3 3200G with Radeon Vega Grahics @3600

GPU: Nvidia GTX 1070 w/ driver version 440.59

Mem: 8GiB DDR4

 

 

I can't help with the handbrake container, but please use our plex container to get it working there first. If you have plex pass that is.

Link to post
8 minutes ago, tmchow said:

Have you installed Nvidia unraid already? Not just the plugin but the modified OS?

The plugin is how the modified OS gets installed. The plugin has no other purpose.

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