1 hour ago, blinside995 said:

Updated to the RC 6 and seems to still be having the same behavior. 

On boot the plugin runs this command.

nvidia-smi --query-gpu=gpu_name,gpu_bus_id,gpu_uuid --format=csv,noheader | sed -e s/00000000://g | sed 's/\,\ /\n/g' > /tmp/nvidia

and outputs to a file /tmp/nvidia

When you go to the plugin page it reads the contents of that file to get the GPU info.

cat /tmp/nvidia

I can't tell you why your copy of that file is getting deleted, but that's what I would bet is causing it.  Try running the above command and see what the output is, both from a fresh boot and after the problem occurs.

 

My guess is you have some cleanup script or process that is deleting stuff from /tmp

 

5 minutes ago, CHBMB said:

On boot the plugin runs this command.

nvidia-smi --query-gpu=gpu_name,gpu_bus_id,gpu_uuid --format=csv,noheader | sed -e s/00000000://g | sed 's/\,\ /\n/g' > /tmp/nvidia

and outputs to a file /tmp/nvidia

When you go to the plugin page it reads the contents of that file to get the GPU info.

cat /tmp/nvidia

I can't tell you why your copy of that file is getting deleted, but that's what I would bet is causing it.  Try running the above command and see what the output is, both from a fresh boot and after the problem occurs.

 

My guess is you have some cleanup script or process that is deleting stuff from /tmp

 

So, running that did bring back all the info for the GPUs. I'll check an see if some script is messing with that. 

26 minutes ago, blinside995 said:

So, running that did bring back all the info for the GPUs. I'll check an see if some script is messing with that. 

If you have anything clearing /tmp regularly, or the kernel is trying to clear memory because you are nearly out, the file would get deleted.
 

 

36 minutes ago, CHBMB said:

 

nvidia-smi --query-gpu=gpu_name,gpu_bus_id,gpu_uuid --format=csv,noheader | sed -e s/00000000://g | sed 's/\,\ /\n/g' > /tmp/nvidia

My guess is you have some cleanup script or process that is deleting stuff from /tmp

Why is this being done in lieu of the plugin calling nvidia-smi via exec?
Surely polluting tmpfs is avoidable in this case?

Just curious, there's probably a good reason I can't think of.

11 minutes ago, Xaero said:

If you have anything clearing /tmp regularly, or the kernel is trying to clear memory because you are nearly out, the file would get deleted.

1

I don't really know what would be cleaning it.. Any ideas why?

 

im pulling my hair out on this one. would there be any reason to think that there maybe any benefit going from 6.6.7 to 6.7.0 rc6. im a newb at this sorry

 

edit: okay im dumb i read others say copy and pasting entered unseen characters but i didn't believe that could be the issue. guess what i typed rather than pasted and it finally worked. Dont be like me i wasted so much time swapping in 3 different gpu's and in the end it was sooo simple and i even read the fix but dismissed it without test.

Edited March 27, 20197 yr by huskycdn

14 hours ago, Xaero said:

If you have anything clearing /tmp regularly, or the kernel is trying to clear memory because you are nearly out, the file would get deleted.
 

 

Why is this being done in lieu of the plugin calling nvidia-smi via exec?
Surely polluting tmpfs is avoidable in this case?

Just curious, there's probably a good reason I can't think of.

It's a couple of factors, firstly I know bash better than php

Secondly, there is a design element to this, with my current approach (which is very similar to what I use for the DVB plugin) the file should be persistent, except in blinside995's case, but we don't know why he's the only one with this issue yet.

So if it's persistent, even if the GPU is used for something else, VM being the obvious thing, the values persist, I was concerned if we tried to do it on every page load then firstly we're polling a GPU that may now be doing something else, risking the integrity of that and secondly we're talking a file of a few bytes here, and Unraid is run entirely from RAM, so as for polluting the /tmp folder, well, that's kind of it's purpose. 

I suppose I could move it somewhere else, but it'd still be in memory, as that's how Unraid works.

12 hours ago, huskycdn said:

im pulling my hair out on this one. would there be any reason to think that there maybe any benefit going from 6.6.7 to 6.7.0 rc6. im a newb at this sorry

 

edit: okay im dumb i read others say copy and pasting entered unseen characters but i didn't believe that could be the issue. guess what i typed rather than pasted and it finally worked. Dont be like me i wasted so much time swapping in 3 different gpu's and in the end it was sooo simple and i even read the fix but dismissed it without test.

Can't say you weren't warned

 

7 hours ago, CHBMB said:

I was concerned if we tried to do it on every page load then firstly we're polling a GPU that may now be doing something else.

That'd be the obvious thing I didn't think of. Yeah, if the GPU got passed through to a VM or something it might vanish from nvidia-smi. 
Always learning.

hi guys, i get this error after i installed nvidia unraid build, " NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running. "

any idea what causing this ? im quite new to server software and hardware.

i have a 1050 ti and tried to install nvidia 6.0.7.rc6 with 418.43 driver verison.

(i used the Unraid Nvidia plugin to install this)

the same error exist when i switch gpu to an older gtx 970.

 

EDIT: i tried to install the unraid nvidia 6.6.6 and with the 410.78 driver

still same error

 

Mobo: Supermicro X9DRi-LN4+/X9DR3-LN4+, Version REV:1.20A

Cpu:   Intel® Xeon® CPU E5-2670 v2 @ 2.50GHz

Edited March 30, 20197 yr by Norrox
more info

2 hours ago, Norrox said:

hi guys, i get this error after i installed nvidia unraid build, " NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running. "

any idea what causing this ? im quite new to server software and hardware.

i have a 1050 ti and tried to install nvidia 6.0.7.rc6 with 418.43 driver verison.

(i used the Unraid Nvidia plugin to install this)

the same error exist when i switch gpu to an older gtx 970.

 

EDIT: i tried to install the unraid nvidia 6.6.6 and with the 410.78 driver

still same error

I have the same error, but by install the unraid nvidia Version it stucks hat checksume passend after thst nothing happend. 

Is there a way to check and see if unRaid even detect that i have a GPU ?

Is there a way to check and see if unRaid even detect that i have a GPU ?

Run lspci -v
This shows you all hardware recognised. You should see your cpu (look for brand name) in the output list and its adress, which you eventualy need if you want to do anything with it.

51 minutes ago, glennv said:

Run lspci -v
This shows you all hardware recognised. You should see your cpu (look for brand name) in the output list and its adress, which you eventualy need if you want to do anything with it.

Thanks, i cant seem to find any reference that my gpu is there? 😮

https://pastebin.com/p1rWMTXT

Edited March 30, 20197 yr by Norrox
to much text

2 hours ago, Norrox said:

Thanks, i cant seem to find any reference that my gpu is there? 😮

https://pastebin.com/p1rWMTXT

It's not there, so either your card is dead, not properly seated or not enough power. Plugged in the power? 

3 hours ago, saarg said:

It's not there, so either your card is dead, not properly seated or not enough power. Plugged in the power? 

 

Ok, now unraid are finding it.. a little embarrassing... i was using the wrong pci-e socket in conjunction with my cpu, my mobo can have 2 and it has 3 pci-e socket per cpu and i was 100% sure that it was in the correct socket, and by chance i saw it when i was glancing over the manual... now i will check and see if i can get the transcoding to work with emby!   

Edited March 31, 20197 yr by Norrox

I have a question, can you do this on an existing install of unRAID or does it require a completely new install? There are no CPU limitations with this are there? I understand on windows you have to have a gen 8 CPU, my unRAID server has Xeon CPU's, will this still work?

35 minutes ago, ashman70 said:

I have a question, can you do this on an existing install of unRAID or does it require a completely new install? There are no CPU limitations with this are there? I understand on windows you have to have a gen 8 CPU, my unRAID server has Xeon CPU's, will this still work?

 

I don't know where you got that info from. 

Install the plugin, download the version you want to use in the plugin, reboot the server and off you go. 

Ok great thank you.

those of you running this; how's the overall quality of doing the hardware transcode?  I have read a few times now that offloading the transcoding may result in poor quality... is there different quality levels based on which card you're using?

 

 

5 minutes ago, dnoyeb said:

those of you running this; how's the overall quality of doing the hardware transcode?  I have read a few times now that offloading the transcoding may result in poor quality... is there different quality levels based on which card you're using?

 

 

Quality levels is set within Emby/Plex, nothing to do with the card.

1 minute ago, CHBMB said:

Quality levels is set within Emby/Plex, nothing to do with the card.

good to know, thanks!  

1 minute ago, dnoyeb said:

good to know, thanks!  

Although by definition, any time anything is transcoded there will be a loss of quality.   Regardless of the hardware that is being used to do that transcoding.

What's the current easiest method to be able to constantly monitor watch nvidia-smi?  Netdata docker?

Not sure if this has been asked or mentioned, I searched for nvdec but was a bit confused, is there a way to get encoding and decoding working in this build? Is it even needed? I saw this post and got curious...

 

https://forums.plex.tv/t/guide-nvdec-hardware-acceleration-patch-for-plex-media-server-on-linux/391322/2

 

Also I'm seeing the GPU memory usage on my P2000 only using like 77MB for normal x264 and 199MB for 4K HEVC, is that normal? Should it be so low. Again I'm new to this so forgive my ignorance, I searched low memory but didn't see any comments.

14 minutes ago, Wavey said:

Not sure if this has been asked or mentioned, I searched for nvdec but was a bit confused, is there a way to get encoding and decoding working in this build? Is it even needed? I saw this post and got curious...

 

https://forums.plex.tv/t/guide-nvdec-hardware-acceleration-patch-for-plex-media-server-on-linux/391322/2

 

Also I'm seeing the GPU memory usage on my P2000 only using like 77MB for normal x264 and 199MB for 4K HEVC, is that normal? Should it be so low. Again I'm new to this so forgive my ignorance, I searched low memory but didn't see any comments.

 

Here you go.  https://gist.github.com/Xaero252/9f81593e4a5e6825c045686d685e2428.  You can put this script in User Scripts so that you can run it each time Plex gets updated.

 

And yes, I'm seeing that GPU memory usage is about what I'm seeing as well.

Edited April 2, 20197 yr by IamSpartacus