flaggart

Thankyou for the suggestions but unfortunately I've tried these ideas. I have tried binhex and official. I have added the claim token inside the docker config and I am signed into plex in the webgui. I have lifetime plexpass.

I am using the official Plex docker. I have tried with and without the kernel parameter. In the plex logs, there are errors stating "Failed to initialise VAAPI connection: -1 (unknown libva error)".

@ich777 Hello again, I wonder if you would be willing to revisit this issue now that the latest beta includes kernel 6.8.12? I have a fresh install of 7.0.0 beta1 on the Framework Intel i5-1240P motherboard. I have installed Intel GPU Top and GPU statistics. The GPU shows as Alderlake_p (Gen12) @ /dev/dr/card0 and is also listed as available in the Plex transcoder configuration. I have tried setting i915.enable_guc=3 as a kernel parameter, but alas Plex will only use the CPU for transcoding. Thanks for any help.

I have tested today unraid with kernel 6.7.5 and still have same issue Intel Alderlake_p (Gen12) @ /dev/dri/card0 shows in intel_gpu_top and in plex settings it is selectable, but it is not used on transcode

That sucks. But thanks for helping.

Tried with 3 versions now. I used the PLEX_CLAIM method, same result.

Thankyou both for your help with this so far. @ich777 I created i915.conf and rebooted. No change. Here is the docker config, I am using linuxserver however I also tried the official docker with the same result: I don't know what the necessary steps for Plex with the claim are. The Plex server webconfig shows it is claimed under my Plex account, which has PlexPass.

Running Plex with a single transcode, hits the CPU, no GPU usage: Visible in Plex: Errors in Plex log: Jan 22, 2024 09:09:53.968 [22679530531640] DEBUG - [Req#367/Transcode] Codecs: testing h264 (decoder) with hwdevice vaapi Jan 22, 2024 09:09:53.968 [22679530531640] DEBUG - [Req#367/Transcode] Codecs: hardware transcoding: testing API vaapi for device '/dev/dri/renderD128' (Alder Lake-P Integrated Graphics Controller) Jan 22, 2024 09:09:53.968 [22679530531640] ERROR - [Req#367/Transcode] [FFMPEG] - Failed to initialise VAAPI connection: -1 (unknown libva error).

Yes, it is in docker config and I am able to select the GPU in Plex web config, but it can never actually initialize or use it

Hello I am unable to get hardware transcoding working at all with an i5-1240P (Alderlake CPU in Framework mainboard). I am able to select the GPU in Plex web config, but it can never actually initialize or use it: Jan 22, 2024 09:09:53.968 [22679530531640] DEBUG - [Req#367/Transcode] Codecs: testing h264 (decoder) with hwdevice vaapi Jan 22, 2024 09:09:53.968 [22679530531640] DEBUG - [Req#367/Transcode] Codecs: hardware transcoding: testing API vaapi for device '/dev/dri/renderD128' (Alder Lake-P Integrated Graphics Controller) Jan 22, 2024 09:09:53.968 [22679530531640] ERROR - [Req#367/Transcode] [FFMPEG] - Failed to initialise VAAPI connection: -1 (unknown libva error). Unraid 6.12.6 Tried linuxserver and official Plex docker Any ideas? Thanks

Hello, I am considering a project with a 12th-gen Framework laptop mainboard - https://frame.work/gb/en/products/mainboard-12th-gen-intel-core?v=FRANGACP06 I currently have a reasonably large Unraid NAS, the main purpose for which is a plex server. This is composed of the Unraid host, which is an unremarkable ATX pc case, and 2x 4U cases for the drives (up to 15 drives in each). The drives in each 4U case are connected to 4 ports on an internal SAS expander, which then connects via two external mini-SAS cables to the Unraid box. The Unraid box has a single HBA with 4 external ports (9206-16e). I have upgraded my framework laptop to AMD, and have the 12th gen mainboard in the CoolerMaster case, waiting for something interesting to do. I am intrigued to use it for Unraid due to low power usage, the intel CPU being good for plex transcoding, and because it has 4x Thunderbolt ports. . To connect my drives, I am considering going Thunderbolt -> NVME -> pci-e by buying 2x thunderbolt NVME adapters, and 2x ADT-Link NVME to pci-e adapters (such as https://www.adt.link/product/R42SR.html, there are lots of variants). I can then place one of these inside each of my 4U cases, and put a 2-port SAS HBA card (e.g. 9207-8i) in each ADT-Link pci-e slot to connect to the drives via the SAS expander. I am not looking for ultimate speed, just ease of connectivity, and ultra low power usage. I see reviews for some Thunderbolt NVME adapters to be quite low, e.g. 1 to 1.5 Gbit/sec, but others approach 2.5, which would be sufficient (across 15 drives this would be 166 MB/sec during Unraid parity check). Interested to see if anyone has thought of doing similar, or anyone who has experience of speeds using Thunderbolt - > NVME -> pci-e or similar. (PS. Why not Thunderbolt to pci-e directly? Well, these options seem less than ideal as most solutions are geared towards external GPU usage, and are either very expensive / bulky, or are less easy to power (e.g., requires PSU connected via ATX 24-pin, assuming the card is high power). Maybe I am wrong and someone can share a good solution for this approach?)

https://www.highpoint-tech.com/rs6430ts-overview 1x 8-Bay JBOD Tower Storage It seems cheaper for me to buy this directly from Highpoint with international shipping than it is to source the older 6418S locally.

I think I was overthinking it. Any hard link created by these apps will initially be on cache, and mover will move onto array disks. Presumably it is clever enough to move the file and then make the hard link on the same disk filesystem. This seems to be the case from looking at the output of "find /mnt/disk1/ -type f -not links 1", and therefore I can just use rsync with -H flag to move files from one disk to another preserving hard links. I think it is unlikely there are any hard links that span different disks but are hard-linked across the array filesystem.

Hello I use the regular tools such as Sonarr, Radarr etc which keep downloaded files and make a hard copy in the final library location. I want to move all data off of one disk in the array to free it up so that it can be used as second parity disk instead. I realize I can't just dumb copy all of the files from that disk to other disks, as this would lose the hard links and take more disk space as a result. I know that the mover script checks files on the cache disk to see if they are hard links, and I assume if they are, it then finds the original file, and instead of moving from the cache to the destination disk, it instead creates a new hard link. My question is - is there an existing tool I can use to empty one of my disks in this way, or will I need to make a similar script which checks each file to see if it is a hard link, and if so find the original file, etc etc... Thanks

Hello This information is really useful and has inspired be to revamp my setup. I wonder if someone could explain why the single port link speed between HBA and expander would be 2400MB/s rather than 3000MB/s (6gbit * 4 lanes = 24gbit or 3000MB/s). Intended setup: 1x 9207-8e in a PCI-e 3.0 8x slot (max 7.877 GB/s) Single links in to 2x RES2SV240 16 disks per expander, all support SATA3 Would this not provide 3000MB/s to each expander, and therefore up to 187.5MB/sec per disk - rather than 2400MB/s and 150MB/s? Thanks