I'm experiencing various problems which I'm seeing reproducably in
both 2.6.31.2 and 2.6.31.4 after upgrading from 2.6.30.8. Most notably, and
scary to me, is that I'm routinely seeing this I/O error message with
different sector numbers whenever I try to use my usb hard drive.
in syslog:
Oct 13 21:55:49 roll kernel: sd 0:0:0:0: [sda] Assuming drive cache: write
through
Oct 13 21:55:49 roll kernel: sd 0:0:0:0: [sda] Assuming drive cache: write
through
Oct 13 21:55:49 roll kernel: sda1
Oct 13 21:55:49 roll kernel: sd 0:0:0:0: [sda] Assuming drive cache: write
through
Oct 13 21:56:20 roll kernel: end_request: I/O error, dev sda, sector
50334647
in messages:
Oct 13 21:56:20 roll kernel: sd 0:0:0:0: [sda] Unhandled error code
Oct 13 21:56:20 roll kernel: sd 0:0:0:0: [sda] Result: hostbyte=0x07
driverbyte=0x00
Now, this doesn't appear to affect anything so far as the filesystem
doesn't even get set readonly. (It's ext4) ... And I just found out that for
some reason the filesystem is set to continue on errors. I'm surprised, I
thought the default for mke2fs was to set remount readonly. *sigh* I'm
changing the settings for the filesystem now.
In any case, these I/O error messages do *not* appear when I'm doing
similar things on 2.6.30.8. Given that this usb drive is my backup hard
drive, I've reason to be concerned - is this an *actual* error with the
physical disk, is something funky with the kernel, or... what?
Before I attempt to debug this further, I'd like to know what the
error messages are being generated by, so I can find out just how concerned I
should be.
I'm going to assume, for the moment, that something is wrong with
2.6.31.2/4's ability to use my usb hard drive and that the backup I have on
it is likely corrupt since I did not have the filesystem set to remount
readonly on errors. I'm going to wipe it and remake the backup with
2.6.30.8.
Please, let me know what's going on here. I need to know what the error
messages mean - is this a physical failure of the hard drive, or what?
Thank you,
Tim McGrath