[WBEL-users] attempt to write beyond end of device

[Syv Ritch]

On Thu, 9 Dec 2004 18:22:12 -0600, Kirby C. Bohling wrote:
> On Thu, Dec 09, 2004 at 02:47:34PM -0800, Syv Ritch wrote:
>
>> Hi,
>>
>> The server froze a couple of nights ago. No reaction to the
>> keyboard, but the scroll lock, num lock and caps lock keys
>> where flashing.
>>
>> The only way to restart was to press the reset button.
>>
> That's a sure sign of a kernel panic (the flashing keyboard
> lights a dead give away).
>
>
>> The only error that I got in /var/log/messages was:
>>
>> kernel: attempt to write beyond end of drive
>> kernel: 03:03: rw=0,want=0,limit=1541437675
>>
> That's a sign that you are writting beyond the end of a
> device.  I've seen that a couple of times when a mistake was
> made with LVM, where I had re-sized the lv to be larger and
> extended the filesystem. Then an error happened and the new
> size didn't get saved.  On the next reboot I got those types
> of errors.
>
> (By any chance did you not cut and paste that?  I can't find
> that exact message.  Grep'ing the kernel source for "attempt.*
> end of" comes up with several similar messages, but not that
> one)

Oops: Did not cut and paste, but the exact message was:
"kernel: attempt to write beyond end of device" and not "end of drive"

>
> Even when grepping for "end of", I can't even find anything
> that looks like it could print that message.
>
> It sure looks like it's a typo away (device vs. drive) from
> being an error out of:
>
> ./drivers/block/ll_rw_blk.c:1221
>
> It sure looks like that error is coming from trying to read
> /dev/hda3 (the major:minor for hda3 is 03:03 which I'm fairly
> sure what the 03:03 in that error message means).  According
> to the comment in the code limit should be in units of 1024
> byte sectors.
>
> If you start doing some reverse engineering, it sure looks
> like it shouldn't have printed that error message.
>
>
>> This looks like at the end of the backup. I have a SCSI tape
>> AIC7xxx The tape drive is almost new [less than 6 month].
>> Last night the same tape drive and same tape did run OK.
>>
>
> It doesn't look like the error has anything to do with the
> tape drive from the error message.
>
>
>> The backup is 45 gig, the tape is 80 gig native.
>>
>
> What software are you using to do the backup?  Are you in fact
> backing up any partition that includes hda3?  Any idea how big
> the partitions are?  What type of filesystem is on hda3.  (On
> all of my machines, that ends up being swap /dev/hda1 is
> /boot, /dev/hda2 is /, /dev/hda3 is swap, and /dev/hda4 is an
> extended partition that all of the other filesystems end up
> on).

I am just using tar w/o compression. 
tar cvf /dev/st0 /* > /var/log/backup 2 > /var/log/backup.err

and I assume that 03:03 was the time of the event, because it's 
usually around that time the normal backup ends.


--
Thanks
syv@911networks.com
When the network has to work