What to do with a bunch of HDDs?

Question

Hard disks

Good day!

Tell me, what can be done with a bunch of hard drives?
A certain amount has accumulated, in a different state of course, but all the workers, let them use magnets - blasphemy.
There is an idea to make a large storage, a kind of home seedbox, but how to make a storage out of them so that all disks are exactly one large disk?
I plan to store data there that is not critical to failures: movies, TV shows, music, country set.

Answer 1

You fill the computer with controllers that expand the number of sata (ide and others, what kind of disks you have), you can have several computers, and physically connect the disks.
Be careful, different connection methods may set different restrictions. For example, USB controllers limit the total speed of the disks connected to them (usually only one disk is allowed) - usb2 up to 30 MB / s, usb3 is approximately comparable to sata (5Gb / s is this), pci give no more than 100 MB / s (133 it seems), pci -e 8 will give a maximum of 8Gbps.
Those. for example, do not try to connect 5 disks to one old pci, their total speed will not exceed 100 MB, which is normal for one disk but fatal for an array.
Now second, test each disk for linear and not only speed, both at the beginning and at the end, plus collect statistics (for example, old disks with a large number of remapped bad sectors can give normal speed in one place but bad in another), write each disk speed at its worst.
Thirdly, collect software raids from disks with approximately equal speeds in the worst case, and even more so not all into one huge one, it makes no sense, but for about two years I have been working on a config (then I dismantled it because I bought normal disks) from old disks of different sizes where it was a two-level cascade raid was assembled, where raid0 was assembled from two disks with striping (this speeds up the final result without redundancy) and then raid5 was assembled using this logical volume and three others, while the sizes of all disks were different, and specially created volumes were used on those disks that are larger, I assembled raid0 from the remnants (more precisely, the one that btrfs offers is simply merging partitions into one, by serial attachment), but due to unreliability, this tail was not used for storage.
Using the linux nbd or iscsi utility istgt (or aoe if it is possible to allocate a separate network cable for each logical disk) or even drbd, you can merge disks from different physical computers.
Important, do not use hardware raids - this is a vendor lock, besides, cheap ones built into the motherboard do not give any bonus, and expensive ones are probably meaningless here.
Instead of dmraid, you can use the btrfs or zfs features, where the raid functions are built into the file system (carefully btrfs raid5 / 6 are declared unstable , I didn’t know, I used it for several years, it seems there are no problems, even when I changed a failed disk, but this is of course not an argument)
You should not create one huge array, moreover, even within a large disk I create several small ones (although of course this should come from your tasks), since serving small ones is much more convenient, for example, changing the raid class (raid1 -> raid5) or adding a raid5 disk and reconfiguring can be done in stages, plus raid rebuild is usually slower than rebuilding (i.e. you back up data from a small one, recreate it with a different configuration, especially when you first had 3tb disks and then you start buying 4..6tb, it's enough to use terabyte chunks to not know grief ...
You can use lvm or leave everything at the mercy of btrfs, it's up to you. I'd rather work with several independent mount points, sorting out symlinks if necessary.

Answer 2

Anything. Are you interested in someone else's opinion?

Answer 3

I have such a file cleaner. Since its creation, I have already changed all the physical disks without losing a single bit.
The operating system was going something like this: Redundant disks without MDRAID . But the data is collected using ZFS recepies . Snapshots are foolproof.

Answer 4

how to make a storage out of them so that all the disks are exactly one large disk?

You need a "smart" piece of hardware like Drobo, they know how to use different disks in the array.
The only thing that is extremely important to remember is that if such a piece of iron falls, or some kind of internal failure occurs, the chances of recovery are very low. As soon as it loses or damages metadata - that's it, your data is a homogeneous liquid slurry.
There are too "smart" proprietary algorithms, and too much twisted to make the device "convenient" for the user. So there is really nothing important to store there.