The volume could be offline, that's for sure, but why the machine ?
I'll note here to others unfamiliar that this is similar to what the Auspex will do.
I forget the exact terms, but an inconsistent FS on an Auspex is placed into a mode where NFS requests to the FS are queued and not acted upon.
Meanwhile, the 'head' can do a normal fsck on the FS. Note the FS is never unexported. After the FS passes the fsck, it can be brought back online, and all queued NFS requests are acted upon.
Obviously this is only one implementation, but I liked it when I first ran into it. I'm sure given enough resources NetApp could produce something similar. The question is where something like this is on their strategic plans.
I've seen this happen on an Auspex, but never yet on a NetApp. What does happen on a running system when there's a FS error? Does the server crash or anything? Do you get a warning?