Can RAID5 drives be defective while passing self-tests?

Question

I have a 4-drive RAID5 in Ubuntu 10.10 with mdadm. One of the disks seems to be problematic. It dropped from the RAID, and I re-added it and it was fine for a while, but now when I re-add it I get a FailSpare immediately. I checked the drive with DOS SeaTools, both long and short tests, and they passed. No SMART problems have been tripped. I ran the Ubuntu short self-test and it passed, but now after trying to re-add it, I can't read the drive's SMART data in Disk Utility, or run self-tests.

Is it possible this drive is defective even though it passes the self-tests, or is there something else wrong here?

Update: It turns out one of the SATA ports on my motherboard was likely defective. I replaced 2 (or was it 3) drives that (most probably) had no errors until I tried another SATA controller port and so far it works fine.

Answer 1

Certainly it's possible. Drives that do a lot of "almost failing" reads or seeks may have timing problems, so while they "pass" the test, they may take a lot longer to provide data to the requesting thread. Since your other drives are fine (ie the problem is most likely the problematic drive) I'd subject it to more rigorous testing than Ubuntu's short test. My personal favorite is SpinRite 6, but there are others equally good. You want something that will exercise the drive vigorously, reading and writing each byte of every sector multiple times. For me that's a level 4 scan with SpinRite. Drives that pass this are usually ready for action. PS: A level 4 scan may take 8 to 12 hours on a moderately large drive (500 G).

First however, reboot. If that doesn't get you back to being able to see the smart data for the problem drive (and all the others) then you can be pretty sure the drive is defective.

Answer 2

One of the SATA ports on the motherboard was defective, as detailed in the edit to my post above:

I replaced 2 (or was it 3) drives that (most probably) had no errors until I tried another SATA controller port and so far it works fine.

Answer 3

Question (paraphrased):

Is it possible this drive is defective

Answer: Yes.

Big red blinking light with siren noises:

"I can't read the drive's SMART data in Disk Utility, or run self-tests."