During the creation of the article series FreeBSD as a server, I actually had a hard drive fail. In such a case, I don't hesitate for long and replace it immediately. My data is more important to me than a possible ‘I'll be fine’, even if the disc was back after a reboot. With ZFS, this has lost its horror.
How did I find out about this? Monit with the ZFS extension sent a well-behaved mail that the pool data was no longer consistent. A quick check with zpool status then revealed the details: ada0 was generating write and CRC errors.
Basically, this can be reduced to a single command:
zpool replace POOLNAME OLDDRIVE NEWDRIVE
The article Setting up hard disks with ZFS has been supplemented with this chapter. Knowing this and, above all, how easy it is, is a great relief for your nerves if the worst comes to the worst.
