Jacon

Still works for me on 7.2.4

I'm upgrading my server and would like to sell this complete to someone looking to stand up their first Unraid server. I've been running this setup since 2022 and it's worked flawless for me. No heavy workload on the GPU or mining, only the occasional transcoding stream from Plex. Supermicro X10SLH-F (upgraded 3.4 BIOS) Intel Xeon E3-1275L v3 @ 2.70 GHz Haswell | 4 cores | 8 threads | 3.9Ghz Turbo | 45 TDP (great for low power consumption) Micron 4x8 GiB DDR3 ECC RAM (32GB total) - 18KSF1G72AZ-1G6E1 Noctua NH-L9i (low-profile) CPU fan MASTERWATT 550 Watt Semifanless Modular Power Supply EVGA GTX 1660 Super GPU (6GB) This WILL transcode several 4K streams This has been running in a rack server Silverstone RM-4400 case in a climate controlled closet. I'll let the entire setup go for $350 + shipping. I can't ship this immediately so there will be a couple of weeks wait time until I can swap everything over. Paypal is preferred.

6.12.8 --> 6.12.11...no issues so far.

Has this plugin been deprecated? It's no longer showing in Community Apps.

I have a single folder on a share that I can no longer access via MacSMB, yet the file manager in UNRAID still displays all of the files and the Docker Container (binhex_QbT) pointed to that folder also appears to be working. However, when I attempt to load a file into the docker via the browser on my mac, Finder hangs and freezes. The issue first occurred when I attempted to move more files into the share folder via Finder SMB. After moving the files, I went into the QbT docker UI but finder wouldn't load the folder needed for the torrent. I get the same error when I try to browse the folder from my iPhone. My phone and Mac FInder just show "loading..." indefinitely. The only way to remedy is to close finder and reboot the mac. I initially thought this was an SMB problem or SMB cache issue but I've tried all of the various tricks including adding the following to the SMB Extras: veto files = /._*/.DS_Store/ aio read size = 1 aio write size = 1 strict locking = No use sendfile = No readdir_attr:aapl_rsize = no readdir_attr:aapl_finder_info = no readdir_attr:aapl_max_access = no fruit:posix_rename = yes fruit:metadata = stream SMB Settings Thread I ran an XFS_repair with -nv on both drives associated with the share, but no errors were found I think. I'm able to load the folder just fine in the UNRAID webUI but it does take a couple of seconds to load. This folder has ~2,100 files in it. I've tried copying the folder contents to a new folder, but the new folder still hangs when accessed over SMB. I also reset all permissions using Docker Safe Perms from 'user' to 'nobody', but still getting the same problem. UNRAID 6.12.6 sassnas-diagnostics-20240210-0111.zip

Any chance a BIOS setting was hardcoded that prohibits the BIOS from reading the card? It could be worth a default reset. I'd also check to see if there are any PCI settings that appear odd.

After my own troubles with a failing GPU, I stumbled up on a GPU-BURN docker. I'd like to see someone build this into a Unraid friendly docker given the very notable rise in Nvidia demand over the last year. Those of us using the Nvidia Plugin will eventually run into trouble with our GPU(s) and having the right troubleshooting tools is paramount, since the existing plugin doesn't support the full suite of Nvidia controls, settings and memtest options. GPU-BURN - great little docker but it's command line only and consumes quite a bit of docker img to run, because of the write process needed to stress test the GPU (as I understand it). There are a lot of great options within the docker and integrating them into a simple GUI before executing the test would be fantastic. The command-line output of the test could also be visualized into the docker page. The only way to current remove this docker is from Advanced View. I've used it several times and it works well. CUDA-MEMTEST - The holy grain would be to Unraid'ify this docker. CUDA-MEMTEST is the official Nvidia method of a memtest but for the GPU. Would love to see this as a docker to compliment the Nvidia Plugin. @ich77 already has NVTOP and with the addition of GPU-BURN and CUDA-MEMTEST, I think this would complete the trifecta of tools and bolster the confidence for users of Unraid GPUs.

@ich777 @alturismo Picked up a used GTX 1660 Super off ebay and swapped the 1070 card. It runs great! Chalking this ordeal up to a dead GTX 1070...it was a 2016 build date after all. That aside, before I installed the new card, I ran a docker version of GPU-BURN and sure enough, the 1070 would not complete the test before throwing a XID: 31 READ_ERROR in the syslog. [Remember that I was getting XID: 31 WRITE_ERRORs when Plex attempted to transcode before]. I had to manually close the docker to free up resources because the 1070 failed and locked up. I installed the new GTX 1660 and ran the same GPU-BURN test for 5 mins. Test completed without error. I would love to see someone build an UNRAID docker/plugin for the GPU-BURN docker or a version of the CUDA-MEMTEST on github.

How about make? make-4.4-x86_64-2.txz (or the latest)

Yeah, I really need 4K HDR transcoding capability with a little future-proofing. I guess I can always exchange a T400 if it isn’t up to the task.

Any opinion on a 1660? It’s Turing based. The T400 looks a little light for being able to 4K transcode 3-4 streams.

I bought it used and it worked great for a few months, but then gave me trouble when I had the plugin set for Auto Update and it downloaded the r515 drivers (back to page 102-104). Previous owner said it worked well but he had it sitting for awhile before giving it to me. Then, all of a sudden it completely stopped working last week. I'll take a look at the T-series cards. Thanks!

This is a MSI GTX 1070 Gaming X which I think comes OC from the factory. I've read a couple forum posts (not here) where factory overclocking can cause these errors in Linux. I want to test that therory so I'm curious if there's a way to de-clock the GPU and memory back to nvidia recommended settings to see if that fixes my problem. I bought this card second-hand for $100 a couple months ago, therefore I want to try and figure this out before hitting the nuclear option and buying a new card, like a p2000. I appreciate your suggestions. Thanks.

Well...I loaded up Jellyfin again for the hell of it, and it also reports back the same XID 31 error for the ffmpeg process. NVRM: Xid (PCI:0000:01:00): 31, pid=29372, name=ffmpeg, Ch 00000008, intr 10000000. MMU Fault: ENGINE GRAPHICS GPCCLIENT_T1_8 faulted @ 0x1_ffd0b000. Fault is of type FAULT_PDE ACCESS_TYPE_WRITE This is either a hardware or driver problem. Plex points to the driver, this forum points to Plex and the nvidia developer forums aren't really helpful. All I can gather from nvidia is the XID 31 error means 1) bad driver or 2) bad application request, but since Plex and JF both throw this error, I lean to this being a driver problem. XID 31: Fifo: MMU Error https://docs.nvidia.com/pdf/XID_Errors.pdf

As info, I added this to Extra Parameters. Still no change...reverts to CPU transcoding. --runtime=nvidia --device=/dev/dri:/dev/dri I posted this over to the Plex forums too. Adding info here for reference or in the event someone else has any idea.