Looks like I spoke too soon.

No, there’s nothing wrong with the system now that it’s upgraded to Oracle’s Solaris 11 Express… other than the fact that the pool is incredibly imbalanced. I moved all the data from my 5 disk RAIDZ to a a striped set of 2TB drives. I then removed the RAIDZ, added two more drives to mirror the stripe, and scrubbed the pool. No errors, so everything transferred OK. Finally, I took the old 750’s, and added 4 of them to the new pool, as a pair of mirrored pairs.

The trouble is all of the data is on the 2TB drives (the original set of pairs), and the 750s are empty. Which further means all I/O transactions are served from that original pair. It’s better than it was before, when all the drives were in a single top level VDEV, but now that I have 4 top level VDEVs, it’s not acceptable.

So, I’m running a bunch of zfs send | zfs recv statements, and they will likely take all day today. While they are running, the pool is pretty saturated in terms of I/O bandwidth as well as IOPS, and I cant actually use any of the services, as I’d be changing the old data, and then I’d have to remember to resync it anyway. So, tomorrow! When, of course, I’m back at work and back into the daily routine, which leaves little time for playing with media.