Eric Sandeen wrote:
>
> "Saunders, Andrew (Magma-da/Reading-UK)" wrote:
>
> > Sep 20 10:00:14 thor kernel: Bad write on page 0xc1822a30
>
> The short, but perhaps unsatisfying answer is that you can probably just
> ignore this error. The code in pagebuf_write_full_page that generates
> it is identical to the code in buffer.c:block_write_full_page - except
> block_write_full_page doesn't print an error. This looks like it's just
> a case of XFS being too noisy.
>
> I'll look at it a bit more, but in the end we'll probably just remove
> the warning message.
I'll be interested in the final answer to this too, since I'm seeing it as
well after moving to the 2.4.9 kernel in an effort to get away from these
error messages:
Sep 20 13:29:41 sdssdp8 kernel: __alloc_pages: 0-order allocation failed.
Sep 20 13:33:06 sdssdp8 last message repeated 364 times
But, that didn't work. At least kswapd isn't chewing up 20-40% of my CPU
with 2.4.9 like it was with 2.4.8.
I guess I'll have to hope that the VM gets fixed in 2.4.10....
*sigh*
Dan
--
Dan Yocum
Sloan Digital Sky Survey, Fermilab 630.840.6509
yocum@xxxxxxxx, http://www.sdss.org
SDSS. Mapping the Universe.
|