Greetings,
While trying to duplicate Randy Hron's "dbench has intermittent hang on
2.6.0-test1-ac2" report, I received quite a few "Buffer I/O error on
/dev/hda8, logical block N" messages. (changing elevators makes no
difference fwiw). I went back to test1, and it spat up a couple of "buffer
layer error" messages and associated traces. Attempting to umount
afterward to run fsck left umount in D state. See attachment.
Rebooting and running fsck turned up nothing wrong.
-Mike
Mike Galbraith <[email protected]> wrote:
>
> Greetings,
>
> While trying to duplicate Randy Hron's "dbench has intermittent hang on
> 2.6.0-test1-ac2" report, I received quite a few "Buffer I/O error on
> /dev/hda8, logical block N" messages. (changing elevators makes no
> difference fwiw).
That's just a gremlinlet. You can delete the offending printk for now.
> I went back to test1, and it spat up a couple of "buffer
> layer error" messages and associated traces. Attempting to umount
> afterward to run fsck left umount in D state. See attachment.
Well that's a worry. Is it repeatable?
Andrew Morton wrote:
>Mike Galbraith <[email protected]> wrote:
>
>
>>Greetings,
>>
>>While trying to duplicate Randy Hron's "dbench has intermittent hang on
>>2.6.0-test1-ac2" report, I received quite a few "Buffer I/O error on
>>/dev/hda8, logical block N" messages. (changing elevators makes no
>>difference fwiw).
>>
>>
>
>That's just a gremlinlet. You can delete the offending printk for now.
>
>
>
>>I went back to test1, and it spat up a couple of "buffer
>>layer error" messages and associated traces. Attempting to umount
>>afterward to run fsck left umount in D state. See attachment.
>>
>>
>
>Well that's a worry. Is it repeatable? . . .
>
Any chance this problem is a consequence of not yet having Sean
Estabrooks partial bvec patch in this person's kernel???
<http://www.ussg.iu.edu/hypermail/linux/kernel/0307.3/0861.html>. Jens
said he applied it on 2003/7/27 so it doesn't seem like this could have
made it into 2.6.0-test1-ac2.
At 03:09 PM 7/30/2003 -0700, Andrew Morton wrote:
>Mike Galbraith <[email protected]> wrote:
>
> > I went back to test1, and it spat up a couple of "buffer
> > layer error" messages and associated traces. Attempting to umount
> > afterward to run fsck left umount in D state. See attachment.
>
>Well that's a worry. Is it repeatable?
Not within 30 retry runs. (elves and gremlins)
-Mike
On Wed, Jul 30 2003, Lou Langholtz wrote:
> Andrew Morton wrote:
>
> >Mike Galbraith <[email protected]> wrote:
> >
> >
> >>Greetings,
> >>
> >>While trying to duplicate Randy Hron's "dbench has intermittent hang on
> >>2.6.0-test1-ac2" report, I received quite a few "Buffer I/O error on
> >>/dev/hda8, logical block N" messages. (changing elevators makes no
> >>difference fwiw).
> >>
> >>
> >
> >That's just a gremlinlet. You can delete the offending printk for now.
> >
> >
> >
> >>I went back to test1, and it spat up a couple of "buffer
> >>layer error" messages and associated traces. Attempting to umount
> >>afterward to run fsck left umount in D state. See attachment.
> >>
> >>
> >
> >Well that's a worry. Is it repeatable? . . .
> >
> Any chance this problem is a consequence of not yet having Sean
> Estabrooks partial bvec patch in this person's kernel???
> <http://www.ussg.iu.edu/hypermail/linux/kernel/0307.3/0861.html>. Jens
> said he applied it on 2003/7/27 so it doesn't seem like this could have
> made it into 2.6.0-test1-ac2.
no not unless Mike is using taskfile + pio, and even then I've never
heard of it triggering.
--
Jens Axboe
On Wed, 30 Jul 2003, Andrew Morton wrote:
> Mike Galbraith <[email protected]> wrote:
> >
> > Greetings,
> >
> > While trying to duplicate Randy Hron's "dbench has intermittent hang on
> > 2.6.0-test1-ac2" report, I received quite a few "Buffer I/O error on
> > /dev/hda8, logical block N" messages. (changing elevators makes no
> > difference fwiw).
>
> That's just a gremlinlet. You can delete the offending printk for now.
I can live with the message, it will remind me that there is still an
issue of some kind. I get it doing other things and on other partitions,
obviously.
--
bill davidsen <[email protected]>
CTO, TMR Associates, Inc
Doing interesting things with little computers since 1979.