I see in the FAQ/manual that beyond 4 drives a second parity drive is recommended.
I just upgraded from split parity. I now have 6x8TB drives. I plan to use 5 of them for data.
Anyone know what would happen if I had a single drive failure? Just some of the data would
be recoverable?
I don't love the idea of what will and what won't be recovered.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Here's the thing about recovery. Say you lost a drive. You replace the failed drive and begin recovery. Every drive is now running at 100% for a day? Two days? Longer? So if you have a drive that was close to giving out, the high stress of recovery could casue it to fail.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
I see in the FAQ/manual that beyond 4 drives a second parity drive is recommended.
I just upgraded from split parity. I now have 6x8TB drives. I plan to use 5 of them for data.
Anyone know what would happen if I had a single drive failure? Just some of the data would
be recoverable?
I don't love the idea of what will and what won't be recovered.
You can have up to 256 data drives with single parity and still recover all data from a single drive failure.
The recommendation is related to the risk that you have a second drive failure before you have succesfully recovered all data from the first failure.
Last edit: Leifi Plomeros 2021-04-03
256! That's incredible. I don't understand how that works. But that's cool. I'll take the risk then. Thank you.
Here's the thing about recovery. Say you lost a drive. You replace the failed drive and begin recovery. Every drive is now running at 100% for a day? Two days? Longer? So if you have a drive that was close to giving out, the high stress of recovery could casue it to fail.
Think of it like this:
All the data on your hard drive is stored as an insanely large number of bits which can be either 1 or 0.
You take the first bit from each hard drive and count how many of them are 1.
If it is an even number then you put 0 in the parity file and if it is an odd number you put 1 in the parity file.
Repeat 8 trillion times for each TB of data.
Then later if you have lost one of the bits...
Simply count how many of the remaining bits are 1. If it matches the parity bit then you are missing a 0 otherwise you must be missing a 1.
If however you have lost two disks it is impossible to know what is missing if you only have 1 parity file. For that you need 2 parity files.