I have a clean raid5 array running on Ubuntu 12.10 with 3 disks. Several months ago, mdadm marked one drive as faulty so I ran an extended SMART test and everything came back fine. I then forced it back into the array and it resynced cleanly. Maybe a week later, the drive dropped out again, so I replaced it.
Things were running fine until about a week ago when a different drive fell out of the array. This time I did a little more research and found that people were having success disabling NCQ on the 'green' drives -- which all of mine are. So I did that with all the drives and things have been running smooth since then (actually a little more than a week).
So now I'm thinking that the drive that I thought died a few months ago may have actually been fine, but that I needed to disable NCQ on it as well.
I'm trying to figure out how to test this so for starters I added the old drive back in as a spare. Then I'd like to mark the replacement from a few months ago as fail, that way the spare (the one that failed months ago) will kick in and the array will resync. Then force the one I marked as fail back into the array as a spare (essentially swapping the two). Then if I can run this way for a month or so, I can be confident that all is well.
If after that time everything appears ok, then I can decide if I want to grow my array or migrate to raid6.
Thoughts? Will swapping the drives in the manner I describe work?