ConnerVT

If you confirmed that the video is actually Direct Play (check in Plex dashboard), there shouldn't be stuttering or high CPU usage. Try some other videos to eliminate it being a file issue. Are subtitles turned on? This will usually cause a file to need to be transcoded. A shot in the dark, but one I've recommended and has solved many peoples' issues - Delete the Plex codec folder from your appdata. Sometimes a corrupted codec can cause issues. Restart Plex, and it will download a fresh set of codec.

My first thoughts are if your main requirement is a virtualization platform, you would be better served by choosing an application which is focused on virtualization. If you already have one you are knowledgeable, go with that. If it is all fairly new to you, Proxmox isn't a bad choice, as there is tons of support and a large user base. Unraid is a great home lab platform. Its roots are in NAS and video server, and over the years has matured into so much more. And it does have a great user community to offer assistance. But a Swiss army knife isn't the best choice if you are predominantly only going to use one of its tools.

It looks as the card you bought is a PCIe x1 slot. Not much info from your link (of course, AliExpress) but my guess is that it also has a multiplier chip, which is where the issue is. It takes a single PCIe lane (data signal from your CPU) and splits it into 4 data signals to your drives. In a desktop PC, where you are typically only accessing one drive at a time, this isn't a big issue. But in Unraid, there are several cases when you are accessing multiple drives simultaneously. Writing to the array (and one or two parity drives) or during a parity check (reading all array drives at the same time) is where this falls on it's face. In both of these cases, the drives tend to fall into sync, where the drives all read/write in parallel with one another. You will see this where the access speed starts off a bit slow, then increases up to (near) the drives maximum capable speeds. With the controller needing to chop up the data one drive at a time, access speeds greatly suffer. The blog you linked is somewhat correct: Marvel chips are well documented as to not play well with Unraid, and multipliers should definitely be avoided. But using SAS HBA have their own issues: Expense, cabling, heat and higher energy usage. Best is to have a motherboard which has enough SATA ports to meet your needs (be aware some share PCIe lanes with NVMe slots, where some SATA connectors are disabled when a NVMe is installed). If you need to add more SATA ports, a PCIe to SATA HBA is a viable option if you select correctly. A PCIe x4 to 4 (or 5) SATA should be problem free. I have a 5 port JMB585 in sy main server and it has been problem free. ASM1164 (without a multiplier) is reported to work as well.

Hmmmm... Interesting. As I wrote earlier, I was not 100% sure of this. The documentation should be updated, as what it currently says is either misleading or just plain incorrect: Anyway, crisis has been averted and the new disk is currently rebuilding. As always, thank you so very much for your support!

I assigned the new 16TB drive and restarted the array. Now Unraid is reporting an emulated drive and is rebuilding it on the new 16TB drive. I know I'm old and my memory isn't always working as it should. But it seems that previously Unraid would display (on Main and Dashboard) that the affected (removed/disabled) disk was being emulated. That the drive just isn't displayed definitely puts people in a panic. Did this change in 6.12 (or even in 6.11, since I upgraded from 6.10.3)? For the current behavior is contrary to what is written in the manual:

Started array again, not showing emulated disk in Main or dashboard. gathered this new diagnostic then stopped array. malta-tower-diagnostics-20231009-0920.zip

Diags and some screenhots from Main and Dashboard. malta-tower-diagnostics-20231009-0847.zip

I'm upgrading an 8TB disk in my array to a larger 16TB (same size/model as my Parity Disk). A healthy array which I have had no errors/issues with. Following the documentation, I stopped the array, unassigned the disk to be replaced and restarted the array. The Disk3 that's being replaced did not come up as emulated as expected, and even at the bottom of Main, at Array Operation, Unraid is reporting "Configuration Valid". What is going on here, and how do I recover from this? Currently the original 8TH is removed and 16TB is in the system. System has been rebooted, no change. I have a free slot to reinstall the original 8TB into the system. I recently upgraded to 6.12.4 from 6.10.3. I have done this upgrade before and have not seen this happen.

Really? I do.

https://docs.unraid.net/unraid-os/manual/changing-the-flash-device/

Both will likely give you the same performance transcoding.

That board is basically the same as I've had in my server for the past couple of years (PCIe x4, 5 SATA ports). I've never had a lick of trouble with it, except for the junk SATA cables that came with it, which I tossed shortly afterward. Thinking about it, other than the case and power supply, it is perhaps one of the few things left from my original server build. 😋

Interesting information. Thank you for this. Counterfeiters are not designing and making their own chips, they buy existing designed memory and controller chips off the shelf. They then assemble them on a circuit board and throw that into a case (which resembles the flash drive they wish to pass as). The memory chips are likely to be lower bin from China. I originally felt this was the weak point, but the deeper into this rabbit hole I get, I am more convinced it is the controller chips which have given us all problems. Be it that the controller is an older design, perhaps less capable, or just not programmed with the correct "special sauce" which makes them work more reliably, it is likely the point of failure.

Simplest is an external drive, USB enclosure, mounted with Unassigned devices. I have two of them. I back up files once a week with a simple script which rsync several folders from my array to the external drive. I use two drives, as I swap them every month or so, and keep one at work. There are several backup utilities in Community Applications. Easy to try them out, and see what you like.

I do the same. It all comes down to how "mission critical" your server is, and your wallet.

I don't believe it is directly a function of the USB 3.0 interface. Rather, drives with USB 3.0 are typically newer technology - smaller process node, with memory cells packed more densely together that generate more heat with less space for that heat to disperse. Newer drives typically are larger capacity than their 2.0 counterpart. You need to compare heat output of drives of the same capacity. It isn't fair to compare a 64GB drive to a 8GB drive.

PNY is likely a good choice these days, as I wouldn't think it to be one of the first choices for a counterfeiter to copy. Samsung and SanDisk have a much greater market share, so more likely to be faked.

This is what I upvoted. I waited until 6.12.4 looked as the final update for 6.12 to update my servers (and I came from 6.10.3).

I've never tried (I have an Nvidia GPU in my server), but looks as recently one user has had some success with a 7900. In the past, Plex transcoding on AMD was very hit or miss, usually not supported by Plex. But things have been changing the past year or so. So the answer to your question is "Possibly".

Firefox is my go to. Searching for "sleep" or "resources" in setting finds nothing. 🤨

Which browser is this?

The ambient temperature around the flash drive is really a minor concern (unless you keep your server in an oven of 60C+). Most flash drives have little thought about cooling - a die in a chip package, soldered to a board and stuffed into a plastic housing. Lowering your case temps by 5C isn't going to matter much, if at all. Where heat is an issue is hammering the flash memory chips with sustained, repeated writes. They just aren't meant for such a duty cycle. The repeated accesses generate heat, which spreads between the silicon semiconductor, the copper interconnects, and the interposers that are used to stack the memory cells together. Each is a different material, which expands and contracts at different rates when heated. The normal use case for an Unraid boot drive is the easy life for a flash drive. Write about 1GB to it once, read from it once when you boot, and write an occasional config file of a few KB.

I'll only offer input on what I know. I haven't had enough hands on time to offer answers into ZFS. The Unraid boot drive doesn't need to be anywhere this size. Probably the sweet spot is around 8-16GB. The whole OS fits in about 1GB. When upgrading, the upgrade process copies most of that to a folder on the flash drive, so now you are about 2GB. Throw in several docker templates and such, you haven't even come near to 8GB, no less 126GB. At boot, the OS and needed configuration is loaded into memory, where all of the OS will run from (with very little ever accessed to/from the flash). My recommendation for a flash drive is to grab a small capacity (8-16GB) drive at your nearby chain drug store (or similar). Several reasons for this: No need for large drive - Unraid drives typically are formatted with vFAT. Larger capacities need you to play partition games. Small capacity = older technology node memory - Tend to not be as fragile to heat/stress (I work in semiconductor manufacturing. Trust me on this one) Less likely to be counterfeit - Lately there has been a number of counterfeit Flash drives flooding the market. Unraid requires the flash have a unique GUID programmed in them, something the major companies always do but counterfeiters usually don't (more work cost money). I'm not 100% sure, but I will go with when you create the drive. When you change the file system, the drive needs to be formatted to that new file system. So no, you will need to move all of the data off the drive before converting. Yes to all three, with different pain points. How you do it affects how long it takes to transfer data. You can connect an external USB (or even attach the drive to a spare SATA port) and access the data with the Unassigned Devices (UD) plugin. UD can read most all file systems, so great for moving data from Windows to XFS (or whatever). When many TB are involved, this is usually the fastest way to transfer data. Some folks leave parity off until they transfer data, others use Turbo Write (all disks spinning/reading all the time) for the fastest transfer speeds. Transferring over your network (and I'll assume Samba) will limit your speed some. You can play your media files from wherever you are copying from (after setting Plex/Plex docker to point to those sources), but that will also impact your transfer speeds, as the disk will be thrashing around, trying to read data for the stream and the disk copy. Many folks run Windows VMs (among others) in Unraid. Like most VM platforms, a little bit of a learning curve, but all of the parts are there. I just recently spun up a Win 11 VM, and have had HASSIO running for some time. Hope this helps answer some of your questions.

Unbind the GPU.