I am currently doing a foreground RAID-10 initialization for 10x20TB drives.
I was curious what else happens, other than just zero-ing out the drives?
It must be doing something more than this, otherwise it wouldn't take a week to initialize.
This is an HBA card and not an onboard RAID.
Normally for mirror-derived RAID levels firmware code or whatever replaces it in case of the software RAID just reads the content of one disk and writes it to the mirrored sibling. After block is written firmware / driver updates the dirty bitmap so RAID knows the region is processed and it should treat the corresponding part of the strip as initialized. RAIDs typically try to accelerate reads by reading part of the strip from one disk and another part from the other sibling one (check Linux MDRAID as a reference), and RAID doesn’t know anything about the file system mounted on top of it, it doesn’t want to return gibberish to the caller, so that’s why initialization. Parity-based RAIDs are more complex to initialize if you care, but that wasn’t your question I guess…
