[lug] SD write timeouts, drive appears to be fine

[Jed S. Baer]

I'm getting some HD write errors in kern.log. At the time of the errors,
the only thing running is rdiff-backup, which writes to the drive in
question. rdiff-backup reports 0 errors on multiple backup runs.

I tried using the find -inum command to locate the file where the write
error is occurring, but it finds no such file. Multiple messages have the
same inode number.

The drive is a Hitachi Deskstar, with a lifetime of 63594 hours. I ran a
SMART extended self-test, and the drive passes that, with no attributes
even close the thresholds.

Here's the sort of messages I'm seeing.

scsi_io_completion: 1 callbacks suppressed
tag#0 FAILED Result: hostbyte=DID_OK driverbyte=DRIVER_TIMEOUT
tag#0 CDB: Write(10) 2a 00 23 fd 64 ff 00 03 40 00
blk_update_request: 1 callbacks suppressed
blk_update_request: I/O error, dev sda, sector 603809023
EXT4-fs warning: 7 callbacks suppressed
EXT4-fs warning (device sda1): ext4_end_bio:332: I/O error -5 writing to
inode 29365231 (offset 5234491392 size 5242880 starting block 75476231)
buffer_io_error: 2558 callbacks suppressed
Buffer I/O error on device sda1, logical block 75476120
Buffer I/O error on device sda1, logical block 75476121
Buffer I/O error on device sda1, logical block 75476122
Buffer I/O error on device sda1, logical block 75476123
Buffer I/O error on device sda1, logical block 75476124
Buffer I/O error on device sda1, logical block 75476125
Buffer I/O error on device sda1, logical block 75476126
Buffer I/O error on device sda1, logical block 75476127
Buffer I/O error on device sda1, logical block 75476128
Buffer I/O error on device sda1, logical block 75476129

I've tried some web searches, an and not found anything more suggestive
than possibly bad cabling, though that seemed to be pointed more at USB3,
and file corruption.

Any suggestions for pinpointing what's causing the problem? I'll probably
run fsck on it later on Saturday or Sunday.

Thanks.