So I've been researching HOW I want to store my data.
As per my other thread, So far, BTRFS has won (In RAID1, since it's the only reliable type).
What I'm having trouble deciphering is how concern over bit rot, and random flips, is:
a. Un-needed (due to firmware CRC etc)
c. Somewhat Paranoid (this is where I'm comfortable)
d. Beyond Paranoid
e. Just pure bullshit.
I read about people fixing significant data on their ZFS pools from bitrot, and "yay thanks ZFS!" threads.
I see similar threads from SnapRAID users, having recovered data from old snapshots that were lost from corruption.
But these could fall into the whole "People report when things go wrong, not when they go right" mentality.
But then I read articles such as this (and there are MANY others, especially on Reddit):
Which make me feel like my "need for checksumming" falls into my 'D. or E." grouping above.
I ALSO read conflicting details on 'Linux Raid 5' and whether or not it corrects errors on the fly, since RAID5 needs a form of sum to calculate data if a disk is lost.
apparently some RAID systems do, and some don't? (checksumm on the fly) I can't find solid details.
Does anyone actually handle priceless data as a profession here, and might know more than one dude with a few months research does?
Keep in mind, most of mine isnt priceless; that which is, is backed up in triplicate, but the rest is still 'valuable'.
EDIT: Also, ECC ram is already present. and has had 48 hours of memtest with ECC enabled, and 48 hours with ECC disabled.
Edited by Master_Scythe, 22 February 2017 - 09:21 AM.