[Plugin] Linuxserver.io - Unraid Nvidia


Recommended Posts

Hello,
 
Am I missing something as I do not see my Nvidia quadro p2000 card in the IOMMU list even though Unraid Nvidia works with no issues and transcodes for plex.  
I do have HVM and IOMMU enabled.
 
I wanted to pass this through to a VM and I did do a search first, but the thread is 60 pages long so i do apologize if I missed something blatantly obvious.
 
Any help is appreciated.  Thanks!
You can't passthrough a card you are using for transcoding in Plex, it's one or the other.

Sent from my Mi A1 using Tapatalk

Link to comment
1 hour ago, bndrillo said:

I have a Quadro 5000, I have tried with 6.8 and I tried rolling back to the 6.7, and it still doesn't seem to be able to communicate with the driver,

 

image.png.2e6bd415e6a2e6fa35c3005d2f33263a.png

Quadro 5000 isn't supported by the current driver. It doesn't have support for nvdec/nvenc either, so it's useless for hardware transcoding.

Link to comment
3 hours ago, saarg said:

Quadro 5000 isn't supported by the current driver. It doesn't have support for nvdec/nvenc either, so it's useless for hardware transcoding.

I have a similar problem, the P5000 and P4000 (one in my server) are supported by this driver version (see snip)?

My 710GT and 1080 both showed up in the Info page, and the 1080 disappeared when it was being used as pass through for my VM, but the Quadro P4000 does not show up on the Info page even tho it is showing up in the system devices page.

Is there a way to get the UUID in a another manor and still try passing it through to the docker? Or will it not work at all if its not detected on the Unraid Nvidia page?

Quadro.JPG

Link to comment
I have a similar problem, the P5000 and P4000 (one in my server) are supported by this driver version (see snip)?

My 710GT and 1080 both showed up in the Info page, and the 1080 disappeared when it was being used as pass through for my VM, but the Quadro P4000 does not show up on the Info page even tho it is showing up in the system devices page.

Is there a way to get the UUID in a another manor and still try passing it through to the docker? Or will it not work at all if its not detected on the Unraid Nvidia page?
Quadro.thumb.JPG.8949e375c66bbe03dada385cd4e2d550.JPG
If you scroll up to the last few of my posts this sort of thing has been discussed.

Sent from my Mi A1 using Tapatalk

Link to comment

i have been having a small problem every 7-18ish days where plex locks up (i think just after a transcode starts) and i need to press the reset button on the system or ssh and send reboot command, i can't do a clean shutdown/reboot, i also tried to disable docker and plex wont shutdown.

 

i have included the diagnostics from the last episode it had. i am using a p2000 and i was using 6.8rc7 at the time it happened but it has also done the same for previous versions of unraid 

iceberg-diagnostics-20191228-1024.zip

Link to comment

Hi!

This plugin is FANTASTIC - thank you!  

 

I'm successfully using it with my dockers, but I can't seem to get a 2nd card to successfully passthrough to VM's.  This plugin wouldn't interfere with passing a GPU through to a VM would it?  To be clear, I have 2 cards in my box - one is setup to work with a docker app, the other is not and is to be passed through.  I can use each card with dockers just fine, and while they show up as passthroughable to VMs, the VM's never boot when they're attached.

 

Any ideas/pointers?

 

Thanks!

Link to comment
4 hours ago, mbezzo said:

Hi!

This plugin is FANTASTIC - thank you!  

 

I'm successfully using it with my dockers, but I can't seem to get a 2nd card to successfully passthrough to VM's.  This plugin wouldn't interfere with passing a GPU through to a VM would it?  To be clear, I have 2 cards in my box - one is setup to work with a docker app, the other is not and is to be passed through.  I can use each card with dockers just fine, and while they show up as passthroughable to VMs, the VM's never boot when they're attached.

 

Any ideas/pointers?

 

Thanks!

Stub the card you want to pass through to the VM so the driver doesn't get loaded.

Link to comment
On 12/28/2019 at 12:39 PM, Defylimits said:

I have a similar problem, the P5000 and P4000 (one in my server) are supported by this driver version (see snip)?

My 710GT and 1080 both showed up in the Info page, and the 1080 disappeared when it was being used as pass through for my VM, but the Quadro P4000 does not show up on the Info page even tho it is showing up in the system devices page.

Is there a way to get the UUID in a another manor and still try passing it through to the docker? Or will it not work at all if its not detected on the Unraid Nvidia page?

Quadro.JPG

Note that the Quadro 4000 (Released 2010)  and the Quadro P4000 (Released 2017) are two entirely different products several generations apart from one another.

Edited by Xaero
  • Thanks 1
  • Haha 1
Link to comment

Hi guys,

 

posted this in the plex forum (lnio) but thought I should place here as well. Installed a quadro p2200 and unraid 6.8 stable and have a crazy memory leak in plex. Never had this before installing this plugin and the p2200 being added. Anyway, I'm attaching the diagnostic file here as well. Any help would greatly be appreciated.

 

ps. I had a sata error but have fixed that by moving cable to another sata connector on the motherboard..no errors with this anymore

unraid-diagnostics-20200101-0804.zip

Link to comment
Hi guys,
 
posted this in the plex forum (lnio) but thought I should place here as well. Installed a quadro p2200 and unraid 6.8 stable and have a crazy memory leak in plex. Never had this before installing this plugin and the p2200 being added. Anyway, I'm attaching the diagnostic file here as well. Any help would greatly be appreciated.
 
ps. I had a sata error but have fixed that by moving cable to another sata connector on the motherboard..no errors with this anymore
unraid-diagnostics-20200101-0804.zip
Memory leak in Plex is not something I can do much about with this plugin....

Sent from my Mi A1 using Tapatalk

  • Haha 1
Link to comment
3 hours ago, drpete said:

memory leak in plex

Are you sure it is plex to blame? I see plex getting killed in your syslog, but that might just be due to OOM deciding to kill plex because something else has filled RAM. I admit I'm not entirely sure how to interpret those parts of your syslog, but I am using LSIO plex and not having this issue.

Link to comment

I know its plex because I gave it 32 gig of memory initially and ran that way for years and now it used all the memory so I limited it to 16 in the parameters and it still uses it up. Not right away. I can run 10 streams and use only 1 gig of memory but as the docker runs longer then a while later I am at 50 percent memory used by plex and then within an hour or two plex has used all 16. if I reboot or stop array and start everything back up same pattern. Hope this helps. These are my two cents but I am not the sharpest tool in the shed

Link to comment
5 hours ago, drpete said:

I know its plex because I gave it 32 gig of memory initially and ran that way for years and now it used all the memory so I limited it to 16 in the parameters and it still uses it up. Not right away. I can run 10 streams and use only 1 gig of memory but as the docker runs longer then a while later I am at 50 percent memory used by plex and then within an hour or two plex has used all 16. if I reboot or stop array and start everything back up same pattern. Hope this helps. These are my two cents but I am not the sharpest tool in the shed

What is your transcode mapped to?

Link to comment
On 12/31/2019 at 12:56 PM, saarg said:

Stub the card you want to pass through to the VM so the driver doesn't get loaded.

Alright, I can't seem to find much on how to stub something.  Anybody willing to point me to a guide or something?  

 

Thanks!

 

Link to comment

it is set to /tmp

 

I had/have "Use hardware-accelerated video encoding" checked in transcoding section. This is in addition to "Use hardware acceleration when available" being checked.

 

Reading thru the beginning of this thread it says this cant do encode. could this be the possible sources of my crashing. Also, I removed the variables added to plex to use gpu and removed the nvidia plugin and am running like I was before graphics card install and it is now 6 hours and no memory leaks or crash. Hopefully check marking the encoding wasnt what did it,dont have much hair left to pull out. lol

 

let me know if the /tmp is a problem. 

 

As always thanks for all time and help, it is greatly appreciated.

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