ConnerVT

Not even close to an expert at this. But a quick Google search of "ASM1153 SATA 3Gb/s bridge" finds many hits going back ro 2017 and beyond, with other ASM chips at 6Gb/s. Makes me think nothing may have changed in the base kernel/drivers of Linux in general.

Two years ago, when I built my first Unraid server, I knew zero about Docker. I now have 18 of them running full time on the server, with another 8 or so utilities that I start up when needed. If you are already comfortable with VM (and systems in general) you will quickly like Docker.

Don't feed the trolls.

Overseerr works well with Sonarr and Radarr for searching TV/movies and requesting content. Beautiful interface and is phone friendly as well.

A fan plugged/wired to a Molex connector is already running at 100%.

You can see the specs for my system below, in my siggy. It currently runs 19 dockers and 1 VM (Home Assistant), all on a very modest 1st gen Ryzen 1500X and 16GB DRAM in a mid-tower case. I've done some torture testing, running a parity check while streaming/transcoding a couple of movies and downloading files, and it didn't miss a beat. So unless you have some known scenario where you are really going to ask for more than your system can handle, one system should suffice. As Vr2lo wrote, I would be more concerned that the network would end up being the bottleneck, especially when serving 4K content.

That's why I mentioned it. Seen a number of people fall into that trap. Worse is when you spend hours troubleshooting why your HD stopped working after installing a NVMe into your system. Don't ask me how I know.

Consumer ITX motherboards will work fine. Note that the number of usable SATA ports on motherboards may be affected by use of a NVMe drive. Some that have 6 SATA connectors may only have 4 that can be used with a NVMe installed. Always check the manual before buying, unless you like surprises and disappointment. The case looks nice (I am a SFF fan for years now). With small cases, heat is always a factor. Much depends on where the system is physically located, how hard you push it, and the components you select. I agree with Lolight - I'd probably go with a 12th Gen Alder Lake CPU with integrated video.

Generally, for containers to communicate with one another will need to be on the same network. Edit: Though, now that I look at my config, I run Plex in Host.

I asked most of these same questions when I set up my server nearly two years ago. I'll share my thoughts, but note they are one man's opinion. There will be many other opinions, maybe as many as there are Plex users. First, I found no easy way to transition from my PC Plex server to the Unraid one, if one wishes to keep all of the metadata, watched history, etc. that is in the Plex appdata. I did a fresh, clean install of Plex on the Unraid server and started anew. I think I created my initial Media share, containing a Movies and a TV folder. In each of those I added one or two movies/series (each in their own folder, using Plex naming conventions) so I had content to create my initial Plex Libraries. I did not have any 4K content at that time, but have since added a 4K Movies and a Music folder to my Media folder. I installed the docker, configured the Plex app, and just let it rebuild metadata. I've added much more media to my Plex server since I moved to Unraid (everything was previously packed into 2 3TB drives on my daily driver PC). IIRC, I think I mounted at least one of those drives temporarily as an Unassigned Device then copied files to where I needed using Krusader. But you could also copy files across your network. Either way, I suggest (at least temporarily) turning cache off for your media share (run Mover to insure no files there, then configure the share), and turning on Turbo Write (this will make transfers to the array faster, but requires all drives be spun up). No matter how you slice it, it will take some time to transfer/copy your media over, assuming TBs of data. As tempting as it is to copy all at once, I would focus on one media type at a time (ie. TV) then move on to the next. So no need or benefit to messing with how you have things configured in the PC server. As to which Docker container to use, it comes down to support. As a general rule, most all containers contain the release from the application's developers. Some may include additional stuff, such as plugins, packages, etc. Myself, I tend to go with linuxserver.io containers (as I use for Plex). They are an active community, and containers are updated weekly and issues addressed promptly. Their official support is on Discord (not my preferred platform). Binhex has many good containers, and supports his work here on the Unraid forum. I would personally avoid the one that comes directly from Plex - Docker seems to be the red-headed stepchild to them. No support from them, either on this forum or their own.

Doh! I assumed the cache was NVMe. Must remember not to post until after I drink my morning coffee. 😆

I know it looks a bit gimmicky, but I put a Icy Box IB-M2HSF-703 on my NVMe. It is a heatsink with a 20mm fan. Idle temps now are 38C. Before installing, it was low 50C at idle, and would get hot enough to throttle down transfer speeds.

I always take a moment to consider the CPU cooler when I need ship/transport a system. Tower cases with a big heat sink and a couple of fans can pull loose and cause damage. You can remove the heat sink and reinstall it later or find some way to secure it for transport.

Sonarr I only use for 1080p TV shows. For movies, I run 2 Radarr dockers - Radar and Radar4K. I set a List in the Radar to mirror requests made in the Radar4K.

A lot of good info provided by @whipdancer Even better advice - Don't real time transcode 4K to 1080p. It rarely is a 100% functional (or enjoyable) experience. With you asking about several simultaneous streams suggests people other than you will be watching video. This means you will have people asking for you to fix issues, or just end up not using what you built. 4K to 1080p is a lot of processing. 1080p to 1080p at a lower bandwidth is fairly easy for modern hardware. Best bet is to have both 4K and 1080p copies on your server, and only serve 4K to devices inside your LAN that can play it directly (or transcode burning subtitles without changing bitrate or resolution). This is what streaming services do - have several different copies of the same content, optimized for the capabilities of the device (and network path) requesting to play it. I do similar on my system. I have separate 4K and 1080p movie folders, and only make the 1080p folder available to non-home Plex users. At home, the 4K media is available. For TV shows, I only do 1080p. If you are really set to go the transcoding path, you will need to drastically rethink your budget. $600 for a GFX and a i7-6700 is not even close to the processing power needed. Much better off taking that money and putting it towards more storage in your array.

I do the same, with my boot USB on a motherboard header cable inside my case. I have a Samsung Fit Plus (a tiny little thing) plugged in outside, for when I quickly want to move some files with a flash drive. Where are my manners? Welcome to the forum. Mostly good folks here. I started my server as a budget build using components I had around the house. The more I play with the server, the more I found for it to do. Somehow, the budget grew in the process.

My server has a similar config (see siggy below). With 18 Dockers running, and 3 of my drives always spinning, it idles around 65W. I doubt messing with C-States will find you much savings vs the time and performance it may cost. Given the list you shared, I would expect that the GTX1050ti, H310 and 10 HDDs contribute the large majority of the 60W you are seeing. Depending how comfortable you are experimenting with your server, you could test my theory by making a trial boot drive and temporarily strip your system down. Remove the GPU, unplug drives, remove HBA. With no Docker/VM running, see how much power you draw with these components out of the equation.

Do you have Plex Pass for your server's Plex account? Plex only does hardware transcoding if you have Plex Pass. Regardless of the OS.

Use the Backup/Restore Appdata plugin to backup the Docker Appdata. (Note when it is backing up, the plugin also stops the Dockers so it can access data, just as Mover needs to. It restarts any of the dockers which were running once the backup is completed). Docker templates for every Docker container you have ever installed are saved on your boot flash drive. In the Docker tab, at the bottom left of the page, you click Add Container, and select the container template you want from the drop down. Any other data (such as media files or whatever) you can back up as you would any data on the system.

Correct. That's why I wrote that twice. 😆 Mover will not move any files that are open (in use by another program/application/system/etc). Even though an individual Docker application (or even all of them) may be stopped, the files are still in open by the Docker Service itself.

You need change "Use cache Pool" from No to Prefer. Then select which of the Cache Pools you wish the Appdata to be stored (looks as you have Cache and Chache 2). You will need to turn off Docker Service then run Mover to move your data from your array to cache. (edit: I see you have already changed from NO to Only. That will work also. Prefer allow if you reach full capacity in the cache drive for data to continue to write to the array. Only will give you an out of space error if you fill the cache drive. In either case, you need to stop Docker Service and run Mover to move data off the array.)

BRIT is correct, IPMI is the integrated solution to all that you ask. WOL really only helps you in one use case - Taking a server which is powered down (or in a sleep/hibernation state) and powering it up to an active state. Even for starting a system which was shut down by a UPS when power goes out, WOL has issues. For you need make the choice - Keep the UPS running on power after the system shuts down, or switch off the UPS as well? If you keep the UPS running, the battery still drains as no computer is fully off. If you turn off the UPS, then the computer won't respond to WOL. Also WOL won't help with hard lock ups - Computer powered but frozen. So these are issues when remote managing standard desktop hardware as a server.

Syslog Server running, saving file to NVMe? (Just spitballing here)

If you are going to run dual parity, both 16TB drives will need be parity drives. The parity drives need to be as large or larger than your largest data drive.

There is no requirement to how many, if any, Parity drives are needed in the array. You can have none, one or two. The only real requirement for the Parity drive(s) is that it must be larger than any of the data drives in the array. If you have 1 Parity drive, you can recover from one drive failing in the array. If you have 2 Parity drives, you can recover from two drives failing. (And as I know someone will chime in shortly - Parity is not a replacement for backing up your data.) The 6 more/less recommendation (the number varies depending on who you ask) is just a threshold of when you may consider a 2nd parity drive. The more drives in your system, the more likely that 2 drives may have an issue. But it makes little sense to have 2 parity drives if only a few data drives, as that makes a larger % of total drive space not available for data. As for Cache, again there is no actual requirement. The original role for the Cache was to be a faster buffer for writes to the array, as the array is generally slower (due to calculating/writing parity). Data sits in the buffer (where it is accessible) and later moved to the array at a scheduled time. So if you do use cache, it should be able to hold the largest file you would write to the array. each defined Share in your system can specify if/how Cache is used. In my system, I do not use cache for my Backup or Media, as they sometimes see very large file sizes and I am in no rush to copy (automated tasks). Those I write straight to the array. Many folks use their cache for other things, such as storing their Dockers' Appdata. In a lot of systems, that cache drive is the only SSD so it is faster access for Docker applications. The recent versions of Unraid allow for multiple cache (now called "Pools"), and these pools can contain multiple drives with some basic RAID modes if desired. So, as you see, you have options. It all depends on what you want the server to do.