Anyway, I still firmly believe that the message is stupid. It’s not even *correct* – in reality there was no device error; there was a checksum error. The most likely reason for that is that the block in question (or its checksum) was written incorrectly in the first place, due to a spot of background radiation or whatever.

Not bothering to think about decent user interfaces and documentation just because it’s targetted at the highly paid is a poor excuse.

The fact that software intended for enterprisey use is almost universally unpleasant to use due to needlessly poor design is another rant…

I’ve had two such drives. I bought an HD103UI when it was on sale on Newegg during some Black Friday hardware purchase binge. It failed within a few months. Found bugs evident in the firmware by looking at it via smartctl (test result reporting and logging was just screwed *up*)), RMA’d drive with attached note about buggy firmware. Got another HD103UI back from them. Ran smartctl on it, noticed same firmware revision. Twigged out on me within a couple months. Attempted to RMA, never heard back.

As for your ZFS error, it was reporting a failed *device*, which is different from filesystem failure. Think of it like a device dropping out of a redundant RAID setup (which ZFS shares some functionality with); yeah, the device died. Throw in a replacement, rebuild (or apparently “zpool replace”, in the case of ZFS), move on.

If you’ve ever used a redundant RAID setup like 1, 5 or 6, or possibly even LVM, you probably got used to a distinction between the filesystem layer and the logical volume the filesystem sat on. ZFS blurs those boundaries by (optionally) using enough redundancy across enough of your physical volumes that, should one fail, you shouldn’t see an operational difference unless you lose more before replacing it.

Interesting parts of that status message: One or more devices has experienced an unrecoverable error. An attempt was made to correct the error. Applications are unaffected.

No offense intended, but you’re a sysadmin, using a filesystem targeted at data centers and people who are paid a hefty salary to focus on keeping servers running. Don’t be surprised if there’s a learning curve.