[mythtv-users] SATA lockups on Aver Revo R3610

[John Drescher]

On Wed, Aug 3, 2011 at 10:42 AM, Ian Oliver <lists at foxhill.co.uk> wrote:
> OK, so this isn't strictly mythtv related but ...
>
> My backend is a Revo R3610 running mythbuntu 9.10. I have my recordings
> on an external SATA drive.
>
> I kept noticing a load of this kind of stuff in the logs -
>
> [96390.385812] ata1: exception Emask 0x10 SAct 0x0 SErr 0x1910000
> action 0xe frozen
> [96390.385821] ata1: irq_stat 0x00400000, PHY RDY changed
> [96390.385829] ata1: SError: { PHYRdyChg Dispar LinkSeq TrStaTrns }
> [96390.385865] ata1: hard resetting link
> [96394.248048] ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 310)
> [96394.261494] ata1.00: configured for UDMA/133
> [96394.261509] ata1: EH complete
>
> I ignored it (yes, I know!) until one day the drive disappeared
> entirely. Only a full power cycle of both the machine and the drive
> brought it back.
>
> I'm now investigating, but wanted to make a backup of the drive first.
> Guess what? Yes, trying to make a backup (just with cp and/or fsync)
> causes the drive to be dropped off the bus within 30 seconds.
>
> A bit of googling suggests that removing the proprietary nvidia driver
> might help. Worth a go? I guess this does more than just the graphics?
>
> This machine is effectively headless unless pull it out of the
> cupboard. Can I just blacklist the driver and reboot or should I change
> xorg.con first?
>

I would use ddrescue with the drive if possible to mirror it to a same
size or larger disk or file on some other storage.

Have you looked at the smart status of the drive for
Reallocated_Sector_Ct, Current_Pending_Sector, Offline_Uncorrectable,
UDMA_CRC_Error_Count or Hardware_ECC_Recovered

The raw values for these should normally be 0. Note: Some drives
encode some of these values so if the number is ridiculously high
ignore it.


smartctl --all /dev/sda | grep -e "Reallocated_Sector_Ct" -e
"Current_Pending_Sector" -e "Offline_Uncorrectable" -e
"UDMA_CRC_Error_Count" -e "Hardware_ECC_Recovered"