Defective drive email warnings coming from different device ids?

[SnakeByte]

Quick question (I hope).

I had been receiving errors such as these from time to time through the security emails that FreeNAS generates (Thanks for these btw!)

Code:

(ada6:ahcich15:0:0:0): WRITE_FPDMA_QUEUED. ACB: 61 f0 b0 80 7b 40 40 00 00 00 00 00

(ada6:ahcich15:0:0:0): CAM status: Uncorrectable parity/CRC error

(ada6:ahcich15:0:0:0): Retrying command

(ada6:ahcich15:0:0:0): WRITE_FPDMA_QUEUED. ACB: 61 38 30 1e 61 40 41 00 00 00 00 00


(ada5:ahcich14:0:0:0): WRITE_FPDMA_QUEUED. ACB: 61 68 b8 be 43 40 c2 00 00 00 00 00

(ada5:ahcich14:0:0:0): CAM status: Uncorrectable parity/CRC error

(ada5:ahcich14:0:0:0): Retrying command


(ada1:ahcich1:0:0:0): WRITE_FPDMA_QUEUED. ACB: 61 08 58 9f 91 40 46 00 00 00 00 00

(ada1:ahcich1:0:0:0): CAM status: ATA Status Error

(ada1:ahcich1:0:0:0): ATA status: 41 (DRDY ERR), error: 10 (IDNF )

(ada1:ahcich1:0:0:0): RES: 41 10 58 9f 91 00 46 00 00 00 00

(ada1:ahcich1:0:0:0): Retrying command

After running smartctl -x on each drive however, only a single drive's log contained all of these errors. Subsequently, after replacing this drive (a drive that passed smartctl -t long), these errors stopped.

How is it that a single drive can be seen as ada1, ada5, and ada6, when I never physically moved the drive? Are these assignments not fixed in BSD/FreeNas?

 

[Ericloewe]

Are these assignments not fixed in BSD/FreeNas?

No. Do not assume that they are. Drives should be identified by their serial number.

 

[SnakeByte]

Ericloewe,

Good to know. Is this a BSD thing? I really need to pick up a book on Linux <-> bsd gotcha's like this. When the vendor saw all of those errors, their first reaction was to send me out a motherboard replacement because of how unlikely it was that three different drives would fail like this.

smartctl -x is great at identifying serial numbers (as is putting a label on the outside of the drive tray), however, how can one reliably tie log entries like those above to the actual physical drive when the errors doesn't report the serial number of the drive (or some other unique identifier?)

 

[Ericloewe]

Ericloewe,

Good to know. Is this a BSD thing? I really need to pick up a book on Linux <-> bsd gotcha's like this. When the vendor saw all of those errors, their first reaction was to send me out a motherboard replacement because of how unlikely it was that three different drives would fail like this.

smartctl -x is great at identifying serial numbers (as is putting a label on the outside of the drive tray), however, how can one reliably tie log entries like those above to the actual physical drive when the errors doesn't report the serial number of the drive (or some other unique identifier?)

Generally, you'd identify the drive (say ada2) and then run smartctl -a /dev/ada2.
ZFS on FreeNAS will also use GUIDs to identify the drives, which you can match to the serial numbers and this matching will be unique.

The ada/da identifiers don't generally change often, but always check them when doing something potentially destructive.

That said, I definitely agree that the mismatch of identifiers is a bit painful and I hope that FreeNAS 10 has some effort put into mitigating this.