On Fri, Apr 01, 2005 at 01:32:31PM -0500, Robert Buels wrote:
> Hi all,
>
> Summary:
>
> On a pretty fast server running debian stable and vanilla linux 2.4.29,
> xfsdump is very slow to write media files and produces "__alloc_pages:
> 0-order allocation failed (gfp=0x0/0)" kernel error messages when
> backing up a 1/3 full 915GB partition, and may be causing the machine to
> hard lock.
>
> Any ideas what the problem could be here?
I'd hazard a guess that your page cache has become filled
with XFS metadata buffers, and the memory they are using is
not being made reclaimable sufficiently quickly to avoid OOM
situations. The CVS tree on oss.sgi.com has a kernel patch
(split-patches/cache_defs) which the XFS code in the CVS tree
uses to push on that memory and free it up much more quickly
when we are getting low on memory.
So, you may have more success with a CVS 2.4 kernel, which is
also at 2.4.29 presently.
cheers.
--
Nathan
|