Two things that catch people right after they build the array: It won't come back after a reboot until you persist it. `mdadm --detail --scan` prints the line you need, append it to /etc/mdadm/mdadm.conf, then run `update-initramfs -u` (Debian/Ubuntu). Skip that and the array only lives for the current boot. Turn on monitoring so a degrading disk actually pages you: `mdadm --monitor --scan` (or the mdmonitor service), aimed at whatever email or pager you already use. On a headless NAS that's the difference between catching it early and losing the array to the one drive that was going. For the level on a Pi: RAID1 is the sane default, fast and simple rebuilds. Go 5/6 only if you need the space and accept the slow, stressful rebuild. Run smartd either way so you see a dying drive before the array does.