bman

+1 for WinAmp. Not sure I need anything fancier... can't quite grasp all the newer stuff with the cluttered interfaces If Plex didn't crash my server I'd be all over it, too for the Chromecast love... one day maybe I'll dig into the cause, or just setup a separate box for it.

Wish I lived near you!

I'm not exactly sure I am understanding the issue completely, but would suggest clearance between the crossbars is nothing to worry about, as long as the heat plate is in full and firm contact with your CPU. If you know the heat plate is touching, and you apply a bit of torque to ensure a solid contact to the CPU, then everything should be stable and useful. If there's any wiggle room between CPU and heat plate, then something is definitely amiss.

I'm not sure there's any basis for complaint on the speed front since most photos are a few dozen MB at best. unRAID uses one or two parity drives just like FreeNAS and ZFS can. You're planning on using your server as a backup, which means you have another copy of all your files elsewhere. If something fails, you fix it and copy over any lost data. The server OS and configuration makes little difference in that case as far as best file protection goes. ZFS (not just FreeNAS) offers features unRAID does not for data integrity and recovery, and unRAID offers some FreeNAS does not. To me it isn't worth the cost of higher CPU resources and requiring same size hard drives in each Vdev (array) to use FreeNAS over unRAID. unRAID requires only that your parity drive(s) be at least as large as your largest data drive, and all other data drives are whatever sizes you have available to use. FreeNAS can deliver more read and write throughput than unRAID because it uses multiple same-size disks together for read & write operations, as opposed to unRAID using one drive at a time. Each drive in a FreeNAS Vdev generally is a member of a specialized pool of data, and none can stand on its own if something really goes awry. unRAID uses separate data disks each having its own intact and mountable file system which can be used without unRAID if/as needed, thus minimizing data loss without getting wild with numerous Vdevs each having their own parity disk(s). So which is better? That surely is a question that each person must answer for themselves. I'd suggest your current unRAID server will not let you down for photos as well (assuming you're using a parity drive or two) as for your video streaming. In my mind unRAID is the value choice that doesn't compromise on data protection.

Having searched briefly for anything that's not rackmount and still holds that many drives, I come up short. As far as I can tell, any other form factor (like Lian-li's PC-D8000) adds extra space but no increase in available drive slots. I'm not sure there's a way around it. There is not a large enough market for a single chassis that holds as many drives as you're looking to house, so your choices are few and expensive.

I'd recommend starting with large drives then. They cost a bit more per TB but having fewer of them to worry about leaves much more room for data capacity expansion as you go forward. And you can do it with fewer HBAs and cables. Not that that amounts to much but it can help As 16TB drives become available in another year to 18 months, the 8TB ones should become nicely priced. If you start out with at least two of them now you can upgrade with more of same later without having to upgrade your parity drive size. Getting to 50+ TB in 8 drives is much nicer than getting there in 22 like I have in one of my servers!

If you anticipate needing more storage than 6TB in future, you could add the 6TB as another drive in your array (after preclearing) and copy from each 2TB drive using rsync, or your choice of other method. You could use rsync -av --progress --remove-source-files /mnt/disk[1-3]/* to /mnt/disk4 to move everything in sequence to disk4. This way your parity is always maintained during the move operation. You can eliminate your 2TB drives from your array after the moving is done without losing parity, if that concerns you; Check https://lime-technology.com/wiki/index.php/Shrink_array for details. When I replace a good drive with a larger one I only erase the smaller drive after the rebuild has been successful. As each drive is its own intact mountable volume in linux you can keep the 2TB drives as backup as long as you need (provided you DO NOT use the --remove-source-files switch I mentioned above, of course).

I always keep at least two copies of my photos, and that's generally using two unRAID servers. I have several thousand on my desktop as well, but that's just a staging area while I finish tagging and thinning them out. unRAID has worked very well or this purpose for me. I just created another share on it for photos (and a backup photos one on the backup server) and it's turned out to be a great repository for digital photo frames and the like as well as keeping them safe. I use Robocopy to keep automated backups of my desktop files. I place the command in a batch file and run it from Task Scheduler. Used to use Second Copy which has all sorts of options but it was overkill in the end and I shied away from it.

No recycle bin by default but there is now a plugin to add this feature. I'm a bit more of and old school user from a time when recycle bins did not exist. I get the purpose, and have found it useful a couple times in the past, but a couple out of hundreds of thousands... I don't trouble myself over these things! My point is unRAID has never been the cause of any data loss I have suffered, which is a primary reason I trust it so much even without using ECC RAM and dual parity. I keep backups of anything I care about, and the rest is transient anyway. 170 drives have failed due to old age, poor quality, etc over the years that I have been using unRAID. It's not from the OS, it's just natural attrition. I have lots of servers I deal with between work and home, and over lots of years, it means lots of failures - many more after the past flooding in Thailand when a slew of drives being sold were somewhat questionable, and proved to be so both before and after their warranties were up. Things have been great the past few years, which makes me happy as I can focus on other tasks! Totally with you on all the readyNAS devices chewing up all those parity drives. It's one of the reasons I decided to go with unRAID almost a decade ago. I compared the costs of buying versus building, and during the process my friend introduced me to unRAID. After I saw the benefits and figured out how to get lots of hard drives in a chassis, I haven't looked back. It's the value proposition for large amounts of storage. I keep getting poked by a colleague who runs FreeNAS and he always says I should check out all the new features. But it's fundamentally different, and I can't even entertain it because it would cost me more. I feel for your IT woes on the corporate side. There are lots of folks who claim they know things about data integrity and safety but have the unfortunate characteristic of not learning from their mistakes and correcting them before it's too late. Management at every level has a role to play in this too, though. When the fit hits the shan management always wants to know who to fire, but plainly that's reactionary. I think prudent managers must take a bit of time to verify things for themselves, even if only once per quarter or so. Problems (human or otherwise) are better avoided/corrected than left until you need a new fan.

Eww, haven't noticed that before (but I stay away from USB HDDs usually as flash disks are much handier). Good tip!

I usually stick with adapters for going from SFF-8087 to SFF-8088 and back, and/or reverse or forward breakout cables (SAS-to-SATA) as required for the situation. I have done long cables out the back of one case and into another, but it really hampers servicing ability because you can't move the cases around independent of each other. You'll find all the adapters you need from eBay or Startech or a decent online computer shop if you haven't any stores nearby. Knowing what you have and what you're going to is all the information you need to get the adapters required. For example if you have an HBA or RAID controller that has SFF-8088 (external 4-lane SAS) connectors and you're wanting to connect 4 physical hard drives to it, you want something like this: https://www.amazon.com/Mini-SAS-SFF-8088-26P-SATA-Cable/dp/B00PKN5NXG assuming it's all in one case. If you have a separate case for your physical drives, use an external SAS cable and an adapter bracket to get the SFF-8087 on the inside of the drive chassis, and breakout from there (e.g. https://www.amazon.com/Dual-SFF-8088-SFF-8087-adapter-bracket/dp/B00DGFIMQK)

Shuck the drives from their shells and add them to your server as internal drives. Or if you mean transferring the data, do it one drive at a time. USB is a shared bus, so any time you transfer files from more than one device simultaneously, all transfers slow down.

The only struggle is large parity drive size increases, like my first jump from 3TB parity to 8TB parity which hurt mentally because I didn't have the extra 5TB of data storage for a couple months. Depends exactly how you're going to use your server, but for me I work with a lot of video files, so I know I'll always be increasing in storage. Any new build now starts with two or three 8TB drives, then I'll drop some good smaller drives in for the extra space if I have them lying around. If not, new 8's it is! I buy enterprise drives with 5 year warranties in hopes I'll get at least 6 years from each of them, and a lot of storage happens in 6 years, I have learned.

I don't run ECC on my unRAID servers (well, one of them at work I do just because that's what was given to me to run with) and I don't run dual parity either. unRAID is stable and really good at keeping data safe, so long as you keep on top of drive failures. The File Integrity plugin is great if you want to be extra certain your files are staying intact as time passes by. It is most useful when you have two copies of your data, however, as it reports failures, not imminent problems. For TV & movies that you can rip again, no worries. Chances are great the file will still play through a few bit shifts, and if it doesn't you've got backups at hand anyway.

I've had an Atto ExpressRAID R380 controller in an Apple Xserve connected to a Supermicro 24-bay SAS JBOD chassis running for years. The JBOD chassis has its own power supply, and uses slot adapters to get from external to internal SAS cable connections. The only "rule" I concern myself with is that all my power connections are properly grounded, which means testing the IEC connection cables, as some of them are faulty even from new.

I'm handy with a soldering iron so I'd just add another 8-port connector from old PSU bits I have lying around if it was needed - but that's just me, and also assuming there were no power requirements that would blow up my power supply by performing such a mod

most of the LSI 8-port controllers use 2 4-port SAS connectors on each, so 3 of those would fix you up for all 24 drive bays. The Dell PERC H2xx or H3xx controllers are affordable via eBay and you just need to flash them to IT mode to allow unRAID to use them normally. I've also not had bad luck with the Supermicro AOC-SAS2LP-MV8 HBAs which are ready to use out of the box, but some folks have had some issues. I'm running a 24-drive server at the moment using 3 of those and six SAS cables. It beats 20 or 24 SATA cables which I have in a couple of other unRAID builds.

Nice motherboard! I got the non-IPMI version for a recent build that I haven't quite gotten around to yet. In my previous experiences, though, I have learned to connect ALL motherboard power connectors no matter how many there are. I've had it happen before where I failed (or thought I wouldn't need that extra 12V feed) only to scratch my head a year later when a new add-on card wouldn't work, or some other silly problem I could have avoided. Each 12V wire and each circuit trace on the motherboard are designed to carry only so much current. If there are extra power connectors, it's probably because in some situations the extra current will need to be available to certain slots on the board. Plug in whichever connectors fit. They're keyed so you can't mess them up, no matter how they're labelled.

Where I am the Supermicro 847E2C-R1K28JBO chassis (which includes two 1280W power supplies, and SAS backplanes and front+rear hot swap bays for 44 3.5-inch hard drives) costs the same as 6.37 10TB enterprise (5-year warranty) hard drives. That's 50TB of parity-protected data for your rendering needs. I've never seen a chassis as large as the one you're after for sale in the used market (eBay or otherwise, yet -- I am sure I will one day!) so as far as I am concerned you're buying a new chassis at full price. I don't see how you're going to fit it into such a small budget. Best forward-thinking logic is as already suggested: Spend money on larger drives so you can use cheaper, smaller chassis, like one of the ones you already have... unless you can barter a deal with a good metal bender in your area who can make you what you need.

650W will work fine for what you're up to. Looks like your load will put that supply in its highest efficiency zone, so I'd say you're good to go!

I'm a very happy unRAID customer, having purchased quite a few Pro licenses and a few Plus licenses for different uses. In my years of experience with unRAID, I've not lost any data except by my own silly pressing of the DELETE key when I shouldn't have. I've not had that kind of success with any other RAID solution, free or commercial, hardware or software. To be honest I haven't even had that same reliability with tape backup solutions I have tried, but that's another story. They all have their flaws and best use cases, but for long term reliability unRAID has proven its worth to me more times than I can count. Considering the cost of unRAID licensing from that perspective, it's well worth it. I've thrown out approximately 170 dead hard drives in the years I've been using unRAID. You can do the arithmetic from there. unRAID is still solid, while the drives are just fodder for the melting pot.

I sometimes move 2TB at a time to an archive server I have at work - so I have a 2TB SSD for cache (and a 10GbE connection). If you're not moving more data than your cache device(s) will handle, there's no improvement to be offered for you there. If you use your cache drive for other things like docker images or VMs and the like, you need to have enough space for all those uses as well as the largest amount of data you may wish to transfer to your array at a time.

I get write speeds of about 25MB/s on my oldest, slowest unRAID server, and about 65MB/s on the newest without any tweaking or cache use.

Wow. First thank you for the reply, but this is truly a "why didn't I think of that?" moment You're right, my disk 6 has those folders in its root. And they have subfolders and files, which is weird. Regardless, I now know how to fix it. Cheers!

You can use traditional RAID types to get more read or write speed than a single HDD (or SSD) can provide. unRAID offers no such advantage here, except in the case of cache, which isn't the same as the regular data array. I was merely pointing out that you can get around your mentioned "negative" regarding write speeds and compete with traditional RAID configurations if you really want/need to. The 60TB SSD is so far unobtainable, but is reality, and if you get in line you can probably have one sometime this year if you've got loads of cash to part with. You would of course want at least two of them so you could have parity protection, but why stop there?