JonathanM

Using a backup destination internally is not recommended, as the things that threaten your data would effect your backup as well. You could make it work, but it would be better to focus your efforts on something that is at least removable so it can be physically separated from the server most of the time.

RAID controllers aren't recommended, for this and other reasons. Can you source a plain HBA SAS controller from the recommended controller list? That would greatly simplify things.

I know I'm quoting something you didn't say, but I wanted to expand on it. When you add a passive heatsink to a chip, you help moderate the thermal cycling by adding mass. The overall cooling effect may be minimal, but absorbing the peaks and valleys helps the chip cope. The huge spikes in temp when nvme drives are writing could be bad, it's good to add mass so the spike is spread out, reducing the thermal shock.

Server 1 drwxr-xr-x 18 root root 360 Jan 30 11:03 mnt/ Server 2 drwxr-xr-x 8 root root 160 Jan 5 17:45 mnt/ Server 3 drwxr-xr-x 11 root root 220 Jan 11 08:52 mnt/ Server 4 drwxr-xr-x 9 root root 180 Jan 11 08:54 mnt/ Data points, but no clarity.

That second license thing was from an era when having a USB key die meant you could be down several days waiting on a replacement license file from Limetech. It was meant to allow you to keep a ready backup so you could be up and running instantly vs. waiting. Now that key replacement is online and automated, having a second license unused on standby is no longer really needed. The discounted license was never really MEANT to run a second server, but of course you could. Now, I'm not a Limetech employee, and they surprised EVERYONE with a black friday / cyber monday license sale last year, so anything is possible in the future, but for right now, it's just posted prices.

Unfortunately finding the real manufacturer of any specific brand can be difficult, and big names are not a good indicator of quality. search "gigabyte exploding power supply" for a recent example of what should have been a reputable company screwing the pooch with some abysmal quality control.

Awesome. So many times people say data security is very important, by which they mean their server is the only place that holds their data and they want to make sure it doesn't die. That's great, and it's everybody's goal to keep their server running without data loss, but the answer is always backup, backup, backup.

Did you miss the "up to" before the percent? Different grades and manufacturing quality greatly effect lifespan.

Which means you have a full backup of anything important stored on different physical media somewhere, correct?

Since the words parity and mirror don't belong together when talking about Unraid, perhaps posting a screen capture of the drive list like ChatNoir, and telling use which disks you want to move would clear things up. I'm pretty sure we're on the same page, but "pretty sure" and data loss go hand in hand. Everybody needs to be completely clear on what's going on before you proceed.

Searching the phrase "do power supplies lose capacity over time" provides LOTS of reading material.

Do your cables look like the picture JorgeB posted?

Do you have a connector from the PSU to each drive?

Proper planning eliminates hasty decisions, as trying to do things "quickly" often results in problems. You should always try to keep a minimum free space across the parity array equal to your largest data drive, when you get below that it's time to source another drive and expand the array. You should never be in a hurry to do anything with computers, as proper planning is required to keep your data safe.

Gigabit network will be much faster than copying and reading from a USB drive.

Set up apcupsd on the windows pc, then point the unraid apcupsd to the windows pc. The apcupsd documentation covers all the options you need.

For how long? Typical consumer battery backups are meant to provide enough runtime to safely shut down a machine, or bridge a small gap while a generator is brought online. They are not designed for extended runtime. You can find UPS's rated for extended runtime, many have options for extra battery packs, but they typically cost as much as a well appointed pc.

Research NAT Loopback, hairpinning, etc.

Because you told Unraid to discard the old parity data and start fresh with a new config.

What location path did you search within dupeGuru, and what paths did you map from the host?

Type of RAM is less important than quantity. Does it have 4GB RAM?

Given that a. Each drive in Unraid is independent, so losing multiple drives still leaves all the others readable b. The data you are storing is a second copy I'd be inclined to not run any parity on the backup server. Parity isn't backup anyway, it's high availability so you can continue to use the server while you replace the drive. So keep parity valid on the main box, and don't worry about it for the backup server. Just my opinion. Worth exactly what you paid.

done

Rather complicated scripting. Not something I'm willing to do, but with enough research and educating yourself it is possible.

As you are seeing, it's not a universal thing, my container upgraded without incident.