On Thu, 31 Mar 2005, David Kewley wrote:
> A quick question for anyone who knows: Is 4kstacks still a problem for the xfs
> in kernel 2.6.11? If not, then at what kernel release did it stop being a
> problem?
Funny you should ask, cause there was a thread on this just last week (or
maybe the week before, i forget). The general consensus is that 4k is ok
for light, low intensity (no NFS, LVM, etc) usage, but for anything high
volume, where uptime must be 5 9's, 8k is the safer bet.
>
> RHEL 4 has 2.6.9, and none of its patches obviously refer to xfs, so it's
> probably the stock 2.6.9 xfs code.
>
> Robin Humble wrote on Wednesday 30 March 2005 22:13:
> > On Wed, Mar 30, 2005 at 09:23:56PM -0800, David Kewley wrote:
> > >As you may know, RHEL comes without xfs enabled in the kernel. My
> > >understanding is that RH chose this path because a) they have in-house
> > >expertise in ext3 but not for xfs, and b) they believe that xfs doesn't
> > > offer any advantages to their customers that ext3 cannot provide.
> >
> > a) fair enough
> > b) I think RedHat are wrong
>
> I wonder what they'd say to raw performance numbers. Maybe something like
> "Those differences don't matter in real life." or "Sequential I/O isn't
> representative of real use." :) Whatever the case, the more of their
> customers make the case to them for supporting xfs, and the more that provide
> good hard reasons why, the more likely they are to consider investing in
> in-house xfs expertise.
>
>
> Hey Robin,
>
> Thanks very much for your experience reports & patches. That's very helpful
> to me. I took a gander at your website, and it sounds like we're in similar
> situations. I'm a sysadmin for a computational geophysics beowulf. The
> fileserver I'm asking about is a 9.6TB raw (24x400) 3-ware 9500 based box. :)
>
> David
>
>
--
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Lonni J Friedman netllama@xxxxxxxxxxxxx
LlamaLand http://netllama.linux-sxs.org
|