Well I can tell you from experience no amount of change documentation and preparation can help a Sysadmin doing work after hours (meaning after a full day of work and then some) not to make some stupid mistake like popping the wrong drive out, overworked anyone ?
I've had some worse experience where on IBM ServeRAID controlers when you put in a replacement drive while the array is in 'degraded' mode some awful engineer at IBM decided that if the new drive were to fail while rebuilding the array to a consistent state that they should fail the whole array instead of just warning that the rebuild failed and asking for another replacement (a good one, yeah it seems they sometime ship failed drives as replacement, another IBM genius I would guess). Now everytime I replace drives on IBM servers i'm allways anticipating server rebuilds, but that never gets to the managers or else like it was said earlier in the thread they would have me do this 'maintenance' on christmas eve or some other crazy maintenance window.
BTW: For consistency's sake you can actually boot a floppy to reenable a 'failed' IBM harddrive and set its status to ONL(ine) although this is hardly documented anywhere and thus save yourself from a complete bare metal disaster recovery on IBM servers.