[Plugin] Nvidia-Driver

[Jase]

24 minutes ago, Peuuuur Noel said:

Which plex container are you running? Official, binhex, linuxserver, etc...?

 

Linuxserver and Hotio.

 

24 minutes ago, Peuuuur Noel said:

You access to web client through https://app.plex.tv/ (remote plex web server) or local ip address (web client embedded in container)? Remote and local web clients are distinct and may differ.

 

Both. There is no issue if I access it locally, but, if my users or I access my server via the web through app.plex.tv or my own domain s1001 and s1003 (Network) pop up.

 

24 minutes ago, Peuuuur Noel said:

Have you tried to start from a clean plex appdata folder? Stop the container, rename/move/backup plex appdata folder, start it again, claim your server (if needed), create a new library for the test and test a playback.

 

Yes - the same issue.

 

24 minutes ago, Peuuuur Noel said:

To rollback to your previous install, put back your old plex appdata folder.

 

Yes - the same issue.

 

Please browse here. https://forums.plex.tv/t/nvidia-hardware-acceleration-inconsistently-working-with-web-streaming/828463/145

 

Edited July 26, 2023 by Jase

[Peuuuur Noel]

5 minutes ago, Jase said:

Both. There is no issue if I access it locally, but, if my users or I access my server via the web through app.plex.tv or my own domain s1001 and s1003 (Network) pop up.

 

Ok, so if the issue is only on remote web client, I don't think it's drivers issue but a network issue.

Check if you have network error on your browser by using the Inspector (right click on the page, Inspect) and checking connections in the Network tab while playing a file.

[Jase]

9 minutes ago, Peuuuur Noel said:

 

Ok, so if the issue is only on remote web client, I don't think it's drivers issue but a network issue.

Check if you have network error on your browser by using the Inspector (right click on the page, Inspect) and checking connections in the Network tab while playing a file.

 

If you look here you can see the logs from Plex. https://forums.plex.tv/uploads/short-url/1j6zdnttxhvMQJT1oUI1UfEGscF.txt

 

I will give it a go now using the inspector.

 

I have no idea about Web Browser coding but here is a screen shot from your request.

Edited July 27, 2023 by Jase

[Peuuuur Noel]

@Jase I see a lot of ns_binding_aborted (page load has been stop), ns_error_connection_refused (refused by server) and even a CORS error (refused by security policy).

Remote web client is not allowed to access your playback content on your local server due to security policy issue.

[Jase]

24 minutes ago, Peuuuur Noel said:

@Jase I see a lot of ns_binding_aborted (page load has been stop), ns_error_connection_refused (refused by server) and even a CORS error (refused by security policy).

Remote web client is not allowed to access your playback content on your local server due to security policy issue.

 

What you see is exactly the same issue outside with a brand new setup outside of my network. The issue does not exist on my other server not using a NVIDIA graphics card.

 

The details are;

 

OS: Unraid Version: 6.12.3 

MotherBoard and CPU: Supermicro X9DAi E5 Xeon.

GPU: GeForce GTX 1660 Ti

 

https://forums.plex.tv/t/nvidia-hardware-acceleration-inconsistently-working-with-web-streaming/828463/152

 

Edited July 27, 2023 by Jase

[Peuuuur Noel]

@Jase If it was a driver or GC related issue, I think you should also have this same issue on local web client and desktop/tv/mobile app. They use same API calls to your plex server.

 

Just to rule out this possibility, have you tried clearing browser cache or testing in private browsing? (I know it sounds stupid but sometimes it solves connection issue).

 

Like the plex forum moderator on the post you linked below, I also see a client disconnect in your logs.

[Jase]

6 minutes ago, Peuuuur Noel said:

Just to rule out this possibility, have you tried clearing browser cache or testing in private browsing? (I know it sounds stupid but sometimes it solves connection issue).

 

 

Yes - I have tried all and all. For a moment I also thought it might be a Ubiquiti Gateway security issue...nadda. I have two separate servers Old and New on the same network. Yes, I have cleared all, tried on PC's, Mac's and it's only the browsers that fail with the S1001 S1003 errors. My End users also see exactly the same thing. If they stream with 'Maximum or Unlimited or Direct' there is no issue. ON the Apple TV for example, my server is cool and hums along converting with the Nvidia card remotely and locally

Edited July 27, 2023 by Jase

[Peuuuur Noel]

You can try to post the screenshot of the inspector on the plex forum, maybe the moderator will have an idea about it.

[Jase]

Done! https://forums.plex.tv/t/nvidia-hardware-acceleration-inconsistently-working-with-web-streaming/828463/154

[Jase]

Thank you so much for your help @Peuuuur Noel

[alturismo]

6 hours ago, ich777 said:

@alturismo have you heard of that yet?

 

nope ...

 

and @Jase sorry to say, i just made a quick mockup here and nvidia is working fine here ... in chrome browser ... playing live tv as sample (will always transcode), local browser playback

 

 

also 4k hevc transcoding ... remote playback ...

 

the described "network 100x" errors are most common malformatted streams or some other errors, just as note ...

 

i've seen this errors from iptv usage on plex with no proper setup ...

[Jase]

9 hours ago, alturismo said:

nope ...

 

and @Jase sorry to say, i just made a quick mockup here and nvidia is working fine here ... in chrome browser ... playing live tv as sample (will always transcode), local browser playback

 

 

also 4k hevc transcoding ... remote playback ...

 

the described "network 100x" errors are most common malformatted streams or some other errors, just as note ...

 

i've seen this errors from iptv usage on plex with no proper setup ...

 

 

Thank you for setting up a mockup on your server. 

 

I've tried everything that I can think of to get my Plex server up and running again...tried different Nvidia drivers, different developer versions of Plex like Linuxserver, Hotio, Binhex, removing the docker containers, reinstalling and even deleting the whole docker image and recreating it. starting with clean new setups of Plex instances with alternative media on different drives and even switching out Nvidia video cards. 

 

I'm at a total loss to why this is happening. This server has been running fine for years up until a couple of weeks ago. The only thing that changed was updating Unraid to 6.12.3 and updating the Nvidia drivers. 

 

I'm now going to try a brand new Unraid USB OS to see if there are any changes.

 

Thanks again for your help!

 

 

Edited July 27, 2023 by Jase

[alturismo]

1 hour ago, Jase said:

different developer versions of Plex like Linuxserver, Hotio, Binhex,

may just try the official one ... ?

[ich777]

2 hours ago, Jase said:

I'm now going to try a brand new Unraid USB OS to see if there are any changes.

Have you also tried the official build?

[Jase]

3 minutes ago, ich777 said:

Have you also tried the official build?

 

Not yet, I'm currently backing up my system which will take a while. I will give it a go with the official once the backup is finished.  

[Jase]

So, I gave the official Plex app a go and sadly it's the same issue. 

[Fizzyade]

Since upgrading to 6.12 and updating the nvdia I've been getting GPU errors, like the following:
 

[56386.612293] NVRM: Xid (PCI:0000:08:00): 31, pid=17147, name=ffmpeg, Ch 00000008, intr 10000000. MMU Fault: ENGINE GRAPHICS HUBCLIENT_FE faulted @ 0x1002_06807000. Fault is of type FAULT_PDE ACCESS_TYPE_WRITE
[57317.960662] NVRM: Xid (PCI:0000:08:00): 69, pid=27124, name=ffmpeg, Class Error: ChId 0020, Class 0000c1c0, Offset 00001b0c, Data 00008000, ErrorCode 0000000c

I am using Emby in a docker container, it's the only thing that uses the GPU and once the GPU fault occurs, it kills that ffmpeg which then causes Emby to either just stop playing the stream that was being played, or it jumps ahead to the next episode if it's a TV series that is being watched.

 

Nothing else has changed and the machine was 100% stable prior to updating.

 

I know when I was on a previous Unraid release, I had stuck to an older driver and I never had any issues, when I checked the driver being used it was using "latest", so as a test I changed the settings in the plugin to use " v530.41.03" and all the issues have now gone away, it's back transcoding without any GPU faults.

 

 

[ich777]

On 7/28/2023 at 5:07 PM, Fizzyade said:

Nothing else has changed and the machine was 100% stable prior to updating.

Please post your Diagnostics.

Do you have any kind of scripts or patches installed on your system for the Nvidia driver?

 

EDIT: I have now tried this on my test machine and can‘t reproduce that.

[raftrider]

On 7/28/2023 at 10:07 AM, Fizzyade said:

I know when I was on a previous Unraid release, I had stuck to an older driver and I never had any issues, when I checked the driver being used it was using "latest", so as a test I changed the settings in the plugin to use " v530.41.03" and all the issues have now gone away, it's back transcoding without any GPU faults.

I can confirm I had a similar issue with the latest nvidia driver and reverting to v530.41.03 solved almost all issues except for the web player transcoding now just hangs no errors on the browser. Card is P2000

[ich777]

50 minutes ago, raftrider said:

I can confirm I had a similar issue with the latest nvidia driver and reverting to v530.41.03 solved almost all issues except for the web player transcoding now just hangs no errors on the browser. Card is P2000

Please report that on the Plex forums since this is a Plex issue also read the first few posts on that page. i also can’t reproduce that on my test machine.

 

If you report anything here I need your Diagnostics where you had that issue.

[bdm965]

On 2023/7/20 at PM5点57分, ich777 said:

抱歉，我只为我的插件编译以下分支：

• 生产分公司
• 下一个功能分支
• 旧版分支（驱动程序 v470.x）
• 开源驱动程序

 

但是，如果您使用稳定扩散，我假设您在 Docker 容器中使用它，如果我没记错的话，该容器运行 Cuda Toolkit 自己的版本，通常您看到的输出是驱动程序最高支持 Cuda 12.2并且应该向后兼容旧的 Cuda Toolkit 版本。

thx for reply, It's helpful for me.

[zachlovescoffee]

Hello,

Recently after installing power top and tips and tweaks, I noticed that nvidia-smi is constantly pinging various cores of my CPU. And Ia also notice that the GPU is no longer detected by Nvidia-driver or nvidia-smi if I run the command in shell. Any thoughts?

[ich777]

1 hour ago, zachlovescoffee said:

Any thoughts?

Only if you post your Diagnostocs.  

[J05u]

Hello, can you please advice how I can put my Nvidia 4060 to sleep

GPU statistics reports 52w of power usage when not in use, power state is P0

when i try one thing - to enter nvidia-smi -i 01:00:0 -pm ENABLED - i have no devices found

Card is passed through to VM, but currenly vm is off and card no passed through to any dockers

I found scrip on my old server - 

#!/bin/bash
# check for driver
command -v nvidia-smi &> /dev/null || { echo >&2 "nvidia driver is not installed you will need to install this from community applications ... exiting."; exit 1; }
echo "Nvidia drivers are installed"
echo
echo "I can see these Nvidia gpus in your server"
echo
nvidia-smi --list-gpus 
echo
echo "-------------------------------------------------------------"
# set persistence mode for gpus ( When persistence mode is enabled the NVIDIA driver remains loaded even when no active processes, 
# stops modules being unloaded therefore stops settings changing when modules are reloaded
nvidia-smi --persistence-mode=1
#query power state
gpu_pstate=$(nvidia-smi --query-gpu="pstate" --format=csv,noheader);
#query running processes by pid using gpu
gpupid=$(nvidia-smi --query-compute-apps="pid" --format=csv,noheader);
#check if pstate is zero and no processes are running by checking if any pid is in string
if [ "$gpu_pstate" == "P0" ] && [ -z "$gpupid" ]; then
echo "No pid in string so no processes are running"
fuser -kv /dev/nvidia*
echo "Power state is"
echo "$gpu_pstate" # show what power state is
else
echo "Power state is" 
echo "$gpu_pstate" # show what power state is
fi
echo
echo "-------------------------------------------------------------"
echo
echo "Power draw is now"
# Check current power draw of GPU
nvidia-smi --query-gpu=power.draw --format=csv
exit

 

But GPU idle power dropped from 52w to 50w on idle

Can be issue that info is reported incorrectly from plugin?

Edited August 1, 2023 by J05u

[alturismo]

2 minutes ago, J05u said:

Hello, can you please advice how I can put my Nvidia 4060 to sleep

 

 

nvidia-persistenced

 

this should take your card to sleep ...

 

3 minutes ago, J05u said:

nvidia-smi -i 01:00:0 -pm ENABLED

 

may also not the proper command  if your card is the 1st one, this would be ...

 

nvidia-smi -i 0 -pm 1

 

but be aware, this feature is may dropped sooner or later, persistenced is the actual one