So here's how a Maxtor drive eels it way trough detection mechanisms before they go titsup without warning.
I had two 320GB Maxtor harddisks (salvaged from their horrible OneTouch external USB drives with notoriously faulty controllers)
So here is what's inside one of these:
Some custom IDE-2-USB interface
That disk is a regular 3.5" IDE drive.
When the controller fails (as both mine did about at the same time) you can take the disks out, they are still (Maxtorwize) ok.
It was two such disks that became one of the RAID0's in my file server.
I set up in RAID0 on a Medley RAID SiL controller. Suddenly my clients complain about the drive not beeing available.
Just some time before before this, I noticed that one of the drives had trouble syncing speed up to the other drive. I could hear it ticking, like 2-3 times per hour. That's how I got suspicious first. Since then, I proactively moved stuff from the drive to another drive (after a proper checksum check).
I then check out the fileserver with the RAID, and lo and behold, the controller software had issued warnings about a drive. Their SMART status was OK, but still the RAID controller was complaining.
So I am beginning to think that Maxtor drives fakes their SMART status and dies off without any warning.
Right now I am shuffling about as much as I can from the RAID :)
Anyway, after a 3rd reboot of the fileserver, and stopping other clients from accesing that RAID, I seem to be able to move files from it, uninterrupted, locally on the machine. I just hope this will hold until the disk dies. I'll reconfigure the two disks to RAID1 and wait for the first one to fail. At least then I've only lost 320GB of free space and not 640GB of stuff. (That particular RAID was purely a playground and download setup.)
I have a RAID10 setup of WD disks, let's just say I have a lot more confidence in that! Also I have a spare disk for that RAID should it ever become needed !
UPDATE: May 8th 2012:
The RAID10 actually failed! It went better than expected
