Restarting a failed slice

ODS: Restarting a failed slice

Sometimes when a slice or disk fails you can restart it. The man page suggests this is useful if you accidentally pop a live disk out of a running set or whatever. The vast majority of the time attempting this is harmless whether or not it succeeds. If it succeeds you have bought yourself a little more time before you need to replace the device (but probably not much more -- I'd go out and replace the device as soon as possible.)

# metareplace -e $MIRROR $COMPONENT

In our case we had a single slice on one of our disks fail. The rest of the slices were running properly. We restarted the failed device:

# metareplace -e d5 c0t0d0s5

...and this worked. However since the disk had now generated a history of possible failure, we replaced it within the week.