[Plugin] Nvidia-Driver

[ich777]

Just now, G Speed said:

black screen

It sounds like you are trying this in the web browser correct?

 

Please try a native Plex app like Android or iOS, the web app is known to have issues when forcing a transcode and just display a black screen.

[AgentXXL]

18 minutes ago, G Speed said:

I just set plex up to transcode, but when I do it just gives me a black screen.. Nothing happens
If I revert back to direct play everything works fine

Not sure what i'm doing wrong

 

 

 

You're not using the correct driver for the T1000. If you go to the Nvidia site and do a search for the latest driver for Linux x64, you'll find that you need to use the 535.154.05 version.

[G Speed]

23 minutes ago, ich777 said:

It sounds like you are trying this in the web browser correct?

 

Please try a native Plex app like Android or iOS, the web app is known to have issues when forcing a transcode and just display a black screen.

haha, yes I was just testing..
I tried my phone worked perfectly lol
Thanks

[pashashocky]

On 2/16/2024 at 11:19 AM, ich777 said:

But please make sure that you kill nvidia-persistenced before starting a VM, otherwise you will most likely crash your server.

 

So this is something that would be worth adding to the documentation somewhere, I think it's not fully documented. Read the posts below. 

 

USE AT YOUR OWN RISK! THIS IS NOT DOCUMENTED ON PURPOSE. CAN LEAD TO DATA LOSS IF YOU CRASH YOUR SERVER.

Here are the steps I took with the nvidia-driver installed

• Settings -> VM Manager -> PCIe ACS override -> Set to "Downstream"
• Reboot server
• Check Tools -> System Devices and make sure the GPU and it's sound card are in an IOMMU group by themselves
  • 
• Run `lspci -vnn -s 01:00.0` to verify `Kernel driver in use: nvidia`. 
• At this stage I have `nvidia-persistenced` currently running, and made sure it's not being used by any Docker containers explicitly
  • Run `pidof nvidia-persistenced`
  • Run `kill -9 5212` - or whatever the pid is from the above command
• Now we should be good to pass through the GPU to the VM 
  • Make sure to pass through both the VM and its sound card
  • 
• Switch to XML mode on the same page (top right corner)
  • Make sure the GPU is in the same `slot`, add `multifunction='on'` and increment the soundcard function to `0x1`.
  • 
• Start the VM, install nvidia drivers and you should be good to go!

 

Thank you for your help!

Edited February 20 by pashashocky

[ich777]

Just now, pashashocky said:

So this is something that would be worth adding to the documentation somewhere, I think it's not fully documented.

I strongly disagree about adding that to the documentation!

 

This is not how the plugin is intended to be used in any way. For most users this will basically mean that they crash the server and potential data loss and this is a big reason why I don't want to add that to the documentation.

 

Please post your tutorial somewhere other than this thread (maybe in the VM subforums) because I don't want to support that in any way, I made this also clear in the first post of this thread.

 

4 minutes ago, pashashocky said:

• Run `pidof nvidia-persistenced`
• Run `kill -9 5212` - or whatever the pid is from the above command

BTW, this can be simplified with that:

kill $(pidof nvidia-persistenced)

 

5 minutes ago, pashashocky said:

Settings -> VM Manager -> PCIe ACS override -> Set to "Downstream"

For most users this is also not needed, this depends always on the hardware.

[ich777]

8 hours ago, AgentXXL said:

Nvidia site and do a search for the latest driver

BTW you can see all the latest driver versions for the various driver branches from Unix here.

[pashashocky]

7 minutes ago, ich777 said:

Please post your tutorial somewhere other than this thread (maybe in the VM subforums) because I don't want to support that in any way, I made this also clear in the first post of this thread.

 

Sure, understood your preference, and it's likely user dependent - you are right. I'll keep the notes to myself. 

Although I won't be the last having this question and I'd much rather have a documentation section with USE AT YOUR OWN RISK, rather than having to gather information from 5 different sources and having to experiment...

[ich777]

6 minutes ago, pashashocky said:

Although I won't be the last having this question and I'd much rather have a documentation section with USE AT YOUR OWN RISK, rather than having to gather information from 5 different sources and having to experiment...

The issue with this approach is that the plugin is intended for Docker usage and not for Docker and VM usage, this is simply not supported.

 

The tutorial that you've posted is also missing the USE AT YOUR OWN RISK section because users will definitely ask why their server crashed because they forgot to kill nvidia-persistneced or any other circumstances, there are a few other scenarios where you could crash the server btw...

 

9 minutes ago, pashashocky said:

understood your preference

This has nothing with my personal preference to do.

You have to understand if I write that down, someone at some point will create a post and say that he lost data because of that and maybe this was critical data for themselves, TBH all data from a user is critical in my opinion.

 

I simply don't support that and I won't document that because this is not the use case of this plugin.

 

Please copy past the tutorial in the VM subforums because it is related to a VM and not for the intended use case.

You can link to your post here if you want to but this is nothing that I can/will actively support/document.

[joey911]

On 2/18/2024 at 10:18 PM, ich777 said:

Examples and the parameters are mentioned in the second post of this thread.

 

There are always extra steps necessary for Docker, regardless if you are using AMD, Intel or Nvidia.

 

If you want to pass a device through you always have to specify extra parameters.

 

Hi, You answered the wrong question.   My question was ...can you amend the store description to note that addtional setup is required and to see the support form?

 

Quote

This plugin will install all necessary modules and dependencies for your Nvidia Graphcis Card so that you can make use of it in your Docker containers. In the settings/plugins menu you find your UUID of your graphics card(s). You can also choose your preferred version of the Nvidia drivers, by default it will always grab the latest driver version that is availabe for the current Unraid build.
This plugin satisfies installation prerequisites of the GPU Statistics plugin from Community Apps. With both plugins installed you can display Nvidia GPU utilization on the unRAID Dashboard.

 

And add something like

 

Quote

Addition setup is required for Docker. Please see the support forum for details.

 

[alturismo]

On 2/18/2024 at 11:16 PM, joey911 said:

would you please add "additional setup required for Docker, see forums" or something? Took me a while to find the key values and such but once I did it worked like a charm.  

 

1 hour ago, joey911 said:

Hi, You answered the wrong question.   My question was ...can you amend the store description to note that addtional setup is required and to see the support form?

 

 

personal oppinion, while you are looking for a solution to use your dGPU in a Docker, you prolly will land in this support Thread, its always recommended to read 1st and then decide what are the todo's.

 

every plugin & docker also has a link to its support Threads, here you have a 1st class manual on Page 1 ... so actually, a liitle responsibility should be there to look for the matching Thread and the readme.

 

when we start writing everything anywhere ... it ll end up in even more questions as nobody starts thinking for themselves anymore, my personal oppinion.

 

Sad enough you didnt find the linked support Thread

 

sample

 

so rather learn your systems (not personally attached to you, thats a general thing) and howto use it and howto use all the already helping points included instead asking for more and more and more text ... which actually alsmost nobody reads anyhow ... sample, see how many questions are here already about VM usage with this plugin, while this is mentioned in the description, on page 1, on readme, ...

 

i hope you know what i mean, make no sence to add more text, its more or less already to easy to setup and alot dont like to read anymore (more click and fire and forget) anything and rather ask questions which are answered already ...

[matic]

Hi guys!

 

Recently I can't get my Nvidia Driver plugin to download the latest driver, I've already uninstalled the plugin and when I tried to reinstall it hangs at the same prompt:

-----------------Downloading Nvidia Driver Package v550.40.07------------------ ----------This could take some time, please don't close this window!------------

 

I left it like this for hours, the internet connection is ok. The driver is not booted up and installed so I can't use my GPU.

 

Please help!

[matic]

And now I saw this prompt: 

--------------Can't download Nvidia Driver Package v550.40.07-----------------
plugin: run failed: '/bin/bash' returned 1
Executing hook script: post_plugin_checks

 

[valiente]

Also getting a strange response when trying to upgrade from 545.29.06 to 550.40.07:

 

+==============================================================================
| WARNING - WARNING - WARNING - WARNING - WARNING - WARNING - WARNING - WARNING
|
| Don't close this window with the red 'X' in the top right corner until the 'DONE' button is displayed!
|
| WARNING - WARNING - WARNING - WARNING - WARNING - WARNING - WARNING - WARNING
+==============================================================================

----------------Downloading Nvidia Driver Package v470.199.02-----------------
---------This could take some time, please don't close this window!------------

---------------Can't download Nvidia Driver Package v470.199.02----------------

For some reason it tries to download the earliest available version...

[ich777]

21 minutes ago, matic said:

Recently I can't get my Nvidia Driver plugin to download the latest driver

Can you please provide your Diagnostics? On what Unraid version are you?

 

21 minutes ago, matic said:

I've already uninstalled the plugin and when I tried to reinstall it hangs at the same prompt

Have you yet tried to reboot your server?

 

22 minutes ago, matic said:

I left it like this for hours, the internet connection is ok.

Do you have any Unifi network gear or AdBlocking on your network?

 

May I ask where are you located in the world?

[ich777]

17 minutes ago, valiente said:

Also getting a strange response when trying to upgrade from 545.29.06 to 550.40.07:

Diagnostics please?

 

In your case, please remove the plugin, reboot your server and reinstall the plugin.

[valiente]

9 hours ago, ich777 said:

Diagnostics please?

 

In your case, please remove the plugin, reboot your server and reinstall the plugin.

Uninstall, reboot and reinstall worked!

Thanks

[ich777]

1 hour ago, valiente said:

Uninstall, reboot and reinstall worked!

Glad to hear that it's working again.

Only a few users and only sometimes are having this issue but I haven't found the cause...

[matic]

2 hours ago, ich777 said:

Have you yet tried to reboot your server?

Yes

 

2 hours ago, ich777 said:

Do you have any Unifi network gear or AdBlocking on your network?

 No

 

2 hours ago, ich777 said:

May I ask where are you located in the world?

Slovenia, Europe

 

11 hours ago, ich777 said:

Can you please provide your Diagnostics?

 

diagnostics-20240222.zip

[ich777]

1 minute ago, matic said:

Yes

Please open up a terminal from Unraid and execute this command:

wget -qO- https://api.github.com/repos/ich777/unraid-nvidia-driver/releases/tags/6.1.74-Unraid | jq -r '.assets[].name' | grep "nvidia" | grep -E -v '\.md5$' | sort -V | tail -1

after you've run it please post the result here.

 

I'm still going through your Diagnostics but everything seems fine.

[matic]

21 minutes ago, ich777 said:

Please open up a terminal from Unraid and execute this command:

wget -qO- https://api.github.com/repos/ich777/unraid-nvidia-driver/releases/tags/6.1.74-Unraid | jq -r '.assets[].name' | grep "nvidia" | grep -E -v '\.md5$' | sort -V | tail -1

after you've run it please post the result here.

 

 

RESULT:

nvidia-550.40.07-6.1.74-Unraid-1.txz

 

[ich777]

6 minutes ago, matic said:

RESULT:

This is from an Unraid terminal correct?

 

Then it should be no issue for the plugin to grab the driver.

How much space do you have left on your USB Boot device?

 

Can you download this file (doesn't matter if you do this from a Unraid terminal or from your desktop). If yes how fast?

 

I must have to do something with your Internet connection since the user a few posts above was able to install the plugin just fine after a reboot.

 

Some ISPs have sometimes issues with the connection to GitHub and get really slow download speeds but since the API is working and the plugin can grab the correct package to download I really don't know what's going on...

[matic]

1 hour ago, ich777 said:

Can you download this file (doesn't matter if you do this from a Unraid terminal or from your desktop). If yes how fast?

I did in in few mins.

 

p.s. thanks for the fast response and help so far.

[ich777]

7 hours ago, matic said:

I did in in few mins.

From your desktop I assume?

 

Please try to reboot and install the plugin again, I really don't see a reason why it shouldn't work on your Unraid server. Do you have maybe custom DNS servers set or something like that on your server? I really don't know what could cause this on your system/network...

 

There is enough space on your USB Flash drive correct? The driver is about 210MB in size.

[kyoumei]

Hi,

 

When I attempt to install the plugin, it gets stuck on the "Installing nvidia driver step" and in my terminal, I can see the following messages pop up

 

Quote

[  304.153014] ACPI Warning: \_SB.PCI0.PE70.S1F0._DSM: Argument #4 type mismatch - Found [Buffer], ACPI requires [Package] (20220331/nsarguments-61)
[  304.741950] NVRM: GPU 0000:1b:00.0: RmInitAdapter failed! (0x26:0x56:1598)
[  304.741989] BUG: unable to handle page fault for address: 0000000000004718
[  304.742096] NVRM: GPU 0000:1b:00.0: rm_init_adapter failed, device minor number 0
[  304.744150] #PF: supervisor read access in kernel mode
[  304.744633] #PF: error_code(0x0000) - not-present page
[  304.744858] PGD 8000000137c7d067 P4D 8000000137c7d067 PUD 137c7c067 PMD 0 
[  304.745088] Oops: 0000 [#1] PREEMPT SMP PTI
[  304.745321] CPU: 9 PID: 13640 Comm: nv_queue Tainted: P           O       6.1.74-Unraid #1
[  304.745557] Hardware name: VMware, Inc. VMware7,1/440BX Desktop Reference Platform, BIOS VMW71.00V.18227214.B64.2106252220 06/25/2021
[  304.745805] RIP: 0010:_nv012096rm+0x3b/0xb0 [nvidia]
[  304.746598] Code: 83 b0 b7 02 48 8b bb 68 01 00 00 e8 6f f2 5a 00 85 c0 74 0b 48 83 c4 08 5b 41 5c c3 0f 1f 00 44 89 e7 e8 88 64 b1 ff 48 89 c7 <8b> 80 18 47 00 00 83 f8 01 74 38 80 bf e1 07 00 00 00 74 49 80 bf
[  304.747147] RSP: 0018:ffffc90012a77de0 EFLAGS: 00010246
[  304.747446] RAX: 0000000000000000 RBX: ffff8881035de408 RCX: ffff8881035de408
[  304.747728] RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000
[  304.747991] RBP: ffff88817e46e000 R08: 0000000000000000 R09: 0000000000000000
[  304.748269] R10: ffffc90012a77e88 R11: 0000000065d86309 R12: 0000000000000000
[  304.748534] R13: ffff88817e46b000 R14: ffff888101ed36b8 R15: ffff88816690ac00
[  304.748798] FS:  0000000000000000(0000) GS:ffff88881fc40000(0000) knlGS:0000000000000000
[  304.749074] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  304.749352] CR2: 0000000000004718 CR3: 0000000168e52003 CR4: 00000000003706e0
[  304.749630] Call Trace:
[  304.749897]  <TASK>
[  304.750145]  ? __die_body+0x1a/0x5c
[  304.750402]  ? page_fault_oops+0x329/0x376
[  304.750669]  ? do_user_addr_fault+0x12e/0x48d
[  304.750920]  ? exc_page_fault+0xfb/0x11d
[  304.751164]  ? asm_exc_page_fault+0x22/0x30
[  304.751412]  ? _nv012096rm+0x3b/0xb0 [nvidia]
[  304.752138]  ? rm_execute_work_item+0xed/0x130 [nvidia]
[  304.754586]  ? _raw_q_schedule+0x69/0x69 [nvidia]
[  304.755025]  ? os_execute_work_item+0x48/0x88 [nvidia]
[  304.755471]  ? _main_loop+0xf1/0x115 [nvidia]
[  304.755951]  ? kthread+0xe4/0xef
[  304.756178]  ? kthread_complete_and_exit+0x1b/0x1b
[  304.756411]  ? ret_from_fork+0x1f/0x30
[  304.756639]  </TASK>
[  304.756864] Modules linked in: nfsd auth_rpcgss oid_registry lockd grace sunrpc md_mod zfs(PO) zunicode(PO) zzstd(O) zlua(O) zavl(PO) icp(PO) zcommon(PO) znvpair(PO) spl(O) vsock_loopback vmw_vsock_virtio_transport_common vsock ip6table_filter ip6_tables iptable_filter ip_tables x_tables efivarfs af_packet 8021q garp mrp bridge stp llc bonding tls nvidia_drm(PO) nvidia_modeset(PO) nvidia(PO) intel_rapl_msr intel_rapl_common iosf_mbi video crct10dif_pclmul crc32_pclmul wmi crc32c_intel ghash_clmulni_intel sha512_ssse3 drm_kms_helper sha256_ssse3 sha1_ssse3 aesni_intel crypto_simd drm cryptd mpt3sas mptsas mptscsih rapl backlight intel_cstate intel_agp syscopyarea mptbase i2c_piix4 raid_class input_leds sysfillrect intel_gtt sysimgblt intel_uncore led_class fb_sys_fops vmxnet3 i2c_core scsi_transport_sas agpgart ata_piix button ac unix
[  304.758717] CR2: 0000000000004718
[  304.758986] ---[ end trace 0000000000000000 ]---
[  304.777810] RIP: 0010:_nv012096rm+0x3b/0xb0 [nvidia]
[  304.781181] Code: 83 b0 b7 02 48 8b bb 68 01 00 00 e8 6f f2 5a 00 85 c0 74 0b 48 83 c4 08 5b 41 5c c3 0f 1f 00 44 89 e7 e8 88 64 b1 ff 48 89 c7 <8b> 80 18 47 00 00 83 f8 01 74 38 80 bf e1 07 00 00 00 74 49 80 bf
[  304.781742] RSP: 0018:ffffc90012a77de0 EFLAGS: 00010246
[  304.782046] RAX: 0000000000000000 RBX: ffff8881035de408 RCX: ffff8881035de408
[  304.782342] RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000
[  304.782625] RBP: ffff88817e46e000 R08: 0000000000000000 R09: 0000000000000000
[  304.782916] R10: ffffc90012a77e88 R11: 0000000065d86309 R12: 0000000000000000
[  304.783202] R13: ffff88817e46b000 R14: ffff888101ed36b8 R15: ffff88816690ac00
[  304.783493] FS:  0000000000000000(0000) GS:ffff88881fc40000(0000) knlGS:0000000000000000
[  304.783776] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  304.784061] CR2: 0000000000004718 CR3: 0000000168e52003 CR4: 00000000003706e0
[  304.784362] note: nv_queue[13640] exited with irqs disabled

 

I tried to build a diagnostics file, but with the plugin installed/attempting to install, Unraid will crash when trying to run the nvidia-smi command with the same errors as above in terminal. Everything in the Unraid GUI becomes unresponsive until I remove the plugin manually and force a reboot.

 

Any idea what I can do to fix the problem? I am using a 1650 super in Unraid 6.12.8

[ich777]

7 minutes ago, kyoumei said:

Any idea what I can do to fix the problem? I am using a 1650 super in Unraid 6.12.8

You are running into a Kernel panic but that is caused because you are virtualizing Unraid from what I can see on VM Ware correct?

 

I can't give support for virtualized instances since most of the times you have to pass through some special arguments to get it properly to work.