I just had to return and replace the fourth disk from my NAS. Since January my original 3 Segate Ironwolf disks died one after another. Luckily I could always just replace the faulty disk and return it to the vendor. Looks like I got unlucky and got a bad patch of hard disks. One of the replacement disks died after a week of service as well.

In all cases the faulty disk was detected dring a zfs scrub. First SMART started to complain and after a couple of days the hard disk just died. 2 of the 3 ones had a hard disk head crash (scratching noises after spinup and the disk was just completely borked), one of them had IO errors when reading from a certain part of the disk. In all cases SMART complained before the disk crashed.

Shoutout to CoolBlue at this point. They returned all of my disks and replaced them with new ones. If possible right away, when not available via a voucher so I could just order a new one once available.

My main takeaway from this story is: Always have a spare disk, and once a disk starts to misbehave be prepared. Apart from that - zfs is great. In all cases the weekly scrub detected the faulty disk, so there was no silent error. Maybe I could have been faster and immediately run a scrub after SMART starts to detect something fishy, but a weekly scrub at least gives a certain baseline, which is nice to have.