Oh, here's an idea to try, Paul:

Modify Hijack's auto-spindown feature for longer timeouts:
Code:

[hijack]
spindown_seconds=60


The theory here being that with your HUGE memory size, it's possible that the read-ahead is fetching 30 seconds or so of data off one drive, wherein the other one spins down, and then the spun-down drive is accessed and a pause occurs as it spins up again. The default timeout is 30 seconds; increasing it to 60 should eliminate any such worries.

Cheers