MowMdown

I personally do not use one as I use an ARC GPU, but I have setup dozens of intel systems that do have iGPUs and SR-IOV has never once been part of the equation. The most likely explanation is that the system is trying to use the Xe driver instead of the i915 driver which can be fixed without that plugin, but I would need logs/diagnostics to know for sure. My guess is that plugin is forcing the i915 driver to load instead of the Xe drivers.

Sorry I misunderstood the question, the answer is that you dont need SR-IOV and it just should work. SR-IOV has never been a suggestion to get iGPU transcoding working. The drivers are part of the linux kernel as I stated in my first comment here.

SR-IOV takes the iGPU and physically splits it into separate individual addressable gpus for running multiple VMs each with their own GPU. Docker containers do not need separate GPUs as they all can share the same host hardware akin to how you can render multiple plex transocding sessions on the same gpu. Its the same concept of pinning cores for a VM vs docker using the whole host CPU as needed.

I am fairly confident accessing your data over SMB would not cause read errors. If anything this is likely hardware related. I would think that it could be power related, or your SAS card might be overheating. How are you powering the disks and are you using power splitters? Also do you have active cooling on your SAS card?

Nope. the only thing you need to do after installing the Arc GPU is to add --device=/dev/dri to the extra parameters to the docker container template. No plugins required! SR-IOV is only needed for splitting an intel iGPU between separate VMs. Docker containers do not require the host hardware to be physically split between them unlike VMs when you do hardware passthrough.

The GPU drivers are included in the linux kernel, there is nothing needed to be installed. SR-IOV is for splitting the iGPU among VMs and intel GPU TOP is to get statistics and display them on the unraid dashboard.

Yes the A-series ARC gpus are fully supported. Also they all have the same number of media-engines so you wont get any performance difference between an A310 and an A770. Edit: Most modern hardware should support AV1 decoding.

You simply assign the USB drive to an array slot, either parity slot or an array slot, and start the array. Do be aware, if the disk ID changes, you will need to likely rebuild the data onto the disk. Parity can also become invalidated.

I have a question related to this but not dealing with an issue, just trying to learn something. Is it ever considered "normal" during the unraid boot process to see "Device not ready, resetting" messages? Or does this indicate issues with nvme devices dropping offline?

It's a battlemage GPU, you'll need to either use a 3rd party "Thor's Kernel" or wait until unraid is on 6.18 LTS kernel. B-Series Intel GPUs were not added into the linux kernel until around 6.14 and unraid is currently using 6.12. Since unraid sticks to the Long Term Support kernels for stability, your GPU will not function until this is added in, in a later update. The good news is 6.18 was recently promoted to LTS. It is likely (speculative - my opinion) that we see lime tech release unraid 7.3 with the 6.18 kernel, the bad news is OpenZFS needs to update their stuff to support 6.18 before unraid can use it. We wont get 6.18 in unraid until zfs gets added to 6.18 since unraid also ships zfs

What is the explicit purpose of adding a 3rd disk to the existing pool? Adding a 3rd drive to a btrfs mirror is going to add 2TB usable space but any data written beyond 2TB worth of capacity will not be mirrored as the mirror can only support a total of 2TB given this specific drive configuration. IMO this is pretty useless setup. Changing to ZFS will allow you to use RAIDZ1 giving you 4TB of usable space (assuming you are adding an identical 2TB disk) and giving you protection on all 4TB in the event a single disk is lost.

To be honest, I would move everything off of your br0.4 network and put it back on bridge for starters. br0 networking is just asking for trouble. There are too many moving pieces with your setup to try and troubleshoot the exact issue without starting from zero and working outwards. Put Plex on bridge networking mode, remove any/all firewall related configs you may have setup on your router. Disable remote networking in plex settings for now. Your plex devices should be able to locally connect via UNRAID_IP:32400 and you should be able to visit the plex dashboard at `http://UNRAID+IP:32400/web` Once this can be confirmed working then go to the next step: Open 32400 to unraid's IP in the router firewall and enable remote connections in plex using 32400 and check to see that you can connect externally. (make sure you remove any custom access URLs setup from before) If all of this works, and stays working, the next step would be to introduce the reverse proxy. Using the DEFAULT swag nginx proxy config for plex, change nothing except making it so that it knows you want to use your UNRAID IP and port 32400. Now depending on how you want to proceed with ports, my suggestion is to move the unraid webgui ports from 80/443 to something else like 81 and 4430 if these are not already taken. Then you can set your reverse proxy to using ports 80 and 443 directly. (assuming you took my advice) Remove 32400 from your router port forward rules and instead replace it with 443. (your Reverse proxy should be running on bridge network mode using ports 80 and 443 directly) Next, in plex network settings, disable remote connections entirely, and add the custom access URLs from your https:/plex.mydomain.com:443 like you had before. If plex fails to connect via your custom URL you likely have DNS issues or a NAT loopback problem to solve. I personally run a local DNS resolver so my DNS requests to my domain doesn't require an active WAN connection. Adguard and Pihole can be setup to forward "https://plex.mydomain.com" to my local IP 10.0.0.2 instead of hitting a public resolver and then back to my devices.

Need to post diagnostics

Allegedly the docker system is going to be reworked to be more "compose" like at some point in the future. maybe this will change the behavior.

I would be extremely cautious restoring databases if you didn't take a proper database backup using the proper procedures from the database documentation. No that generally is not an issue, appdata gets permissions set from the containers, changing them manually can break containers though.

To my knowledge that is the standard behavior. The sister containers only "rebuild" when the parent container is updated from the docker page and only the docker page.

post new diagnostics, preferably without rebooting. If youre in safe mode, plugins are disabled.

rclone is built into unraid 7, you can use rclone to mount to a webdav. A user-script would be needed to mount/unmout.

if you're using zfs for your cache pool, see this thread: https://forums.unraid.net/topic/194091-720-rc1-sabnzbd-issue/

verify your appdata location is in tact and read/writeable

I still do think the CPU getting hit hard was possibly from the file integrity plugin so maybe check the schedule it was supposed to run and see if that coincides with the time this was occurring on your system. That plugin does go through FUSE which causes heavy reads(lsof) and hash calculations (CPU utilization). It's very plausible.

No worries I was just spitballing anything I could think of. The only thing I remember at some point during one of the unraid updates was they changed the CPU frequency driver which changes the way the CPU changes it's clockspeeds and turbo boosting. Depending on which version you upgraded from this could be the result of that.

The only things I can suggest right now would be to verify the AIO pump is functioning and to double check that you mounted the cooler to the CPU properly and didn't forget to remove any plastic peels that often come on the copper cold plate. Hitting 5.0GHz on a 12900K even on a single CPU core pushes enough wattage to hit really high CPU temps depending on where you are measuring the temp from (the specific sensor). You could also temporarily disable turbo boosting in the BIOS to see what it behaves like. I am not entirely surprised you're hitting 80C if your CPU is hitting 80-90% utilization. I would expect these kinds of temps even with an AIO cooler.

What kind of CPU cooler are you running? 12900K CPUs when they turbo boost can get very hot. Edit, what are your normal idle temps?

According to your average CPU load it's very low: 12:25:51 up 10:44, 0 users, load average: 1.11, 1.13, 1.06 Cores: 24 top also reports your average idle time is 95%: %Cpu(s): 0.4 us, 0.4 sy, 0.0 ni, 95.1 id, 4.2 wa, 0.0 hi, 0.0 si, 0.0 st You're also running a 12900K which is a CPU that is known to run hot. The suggestion would be to run in safe mode for a little while because you do have quite the collection of plugins and any number of them could be causing an issue for you. My guess is the File Integrity plugin was running calculating file hashes which is sort of CPU heavy.