Beercules

One of the biggest reasons for ECC is stray electrons. It entirely possible that a bit was flipped in memory by an electron. This wasn't caused by a hardware failure but was detected by the ECC and corrected. There is an awesome video on this cause but sadly I cannot find it, if I do I will post it here. Found it!

Odd because when you look at the system information in Unraid 7, it clearly says Ubuntu 24.04 Beercules

--device=/dev/dri add the above to your plex docker "Extra Parameters" I was able to get my a770 LE working with Plex, but sadly my motherboard doesn't support rebar and Arc uses Rebar with its media engine, you can see a near 70% drop in performance using arc without rebar for encoding sadly, so I moved to A2000.

Unraid 7 uses Ubuntu 24.04 as its base. I've moved my unraid from i3 7100 to Ryzen 3600 to Xeon E5-2697A v4 and the only issue I've ran into so far is APSM (power management) on samsung NVME throwing a bunch of ACR errors. Otherwise I have moved my unraid between hardware so many times and not once has it caused an issue. Unraid is extremely hardware agnostic meaning that is designed to function properly on any hardware platform or device, regardless of the specific hardware components it uses. Beercules

No Wrong way to setup Unraid. It's designed to be tailored to your needs. Having the 2 x 8TB in the array with 1 set as parity will give you 8tb of usable space with resiliency against a single drive failure. I recommend XFS for the array file system, this way no data on the array is loaded into ARC, just the cache pool. Setting up a pool of 2 x 1TB NVME in a mirror (I recommend ZFS as this will use ARC cache also, which is RAM) this configuration will provide resiliency against a single drive failure. When creating your shares, you can decide if you want the data to be stored on the Array or Cache pool or Both with the mover moving data between them. I would recommend setting your appdata and system shares (default docker / virtual machine shares) to use Cache only and setup your shares to use cache first then array second. (most common setup) Ensure you mover is scheduled frequent enough so you don't use up the cache pool. You will be able to add additional drives to your array with a simple preclear and add but the cache pool won't be able to be expanded easy. Hope this makes things more clear. Beercules.

I know its possible using scripts shown by SpaceInvaderOne that connecting and interacting with a virtual disk is possible in unraid but I would like to have some type of interface to make this easier to manage, also it would be nice to see native support for docker to access these mounted virtual disk. A great example of why I think these would be helpful is in the Case of Plex docker or other media index / library apps, which has hundreds if not thousands of small image files in its appdata, when moving these files to a new pool it can take forever. If the entire file system was a single file such as VMDK(VMware), VHD (HyperV) or IMG(KVM) we could store them directly in a containerized storage file making them fast and simple to move to a new pool. The alternative is to run plex in a VM, but this would not allow multiple apps to access my A2000 gpu. Maybe there is a simple way of doing this but I am not aware. ALL HAIL BEERCULES

I am running a very similar setup, X99-E ASUS with a E5-2697a and T600. I had serious errors happening from APCI, I would hook up a monitor to the video out and watch the console for any error messages and see if the PC is becoming unresponsive even locally.. I would also watch the "Logs" in unraid after you boot. This is where I found most of my APCI errors.. I ended up having to add a parameter to the grub boot config to resolve my issues. Hope this helps!

One thing to note, that if your using file compression like with ZFS, I notice a considerable decrease in performance when changing my power profile from "Performance" to "balanced" profiles, I was getting under 120MB/sec between a lone M.2 to a RAIDZ1 SSD but the second I swapped back to performance profile it went back to the expected 400mb/sec transfer speed. Looking at the clock frequency of my cores, "Performance" parks the cores at their base clock of 2600mhz, but balanced parks them at 1200mhz.

Look in the community apps, there is a mover tuning plugin that allows customization of the mover per share. It even lets you set a threshold for size of files to be moved etc.

Awesome, thank you for the quick response!

https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/11276 Would it be possible to get these changes ported to Unraid, I just upgraded to an A770 from a T600 in my server but its doesn't show any details with GPU stats or Intel_GPU_TOP. Please and thank you!

Sorry for the late reply, it took this long for the issue to show up, the logs have been uploaded and currently the problem is happening. I have also attached a screenshot with the issue in action. How I noticed it, there is an increase in activity right now on the CPU. In addition you can see from the attached top screenshot there is a sprawl in "UPDATE_3" processes. Mover is scheduled for 1st of month, so this doesn't look like a mover issue, but a BRTFS issue. Look forward to a resolution. blackbox-diagnostics-20230213-2208.zip

This is the second time over the past month I have seen this behavior. using "TOP" i am seeing atleast twenty "update_3" process in the unraid interface you see the mover attempt to run flashing the message "BRTFS operation running" Since this process runs forever it greys out the "stop array" or "Stop mover" button every 1/2 second so you cant even attempt to stop it. It also lags out the interface to the point where I cannot browse shares. A full on reset of the server clears all the unwanted processes and we wait for it to happen all over again. Please fix!