Hi Jaroslaw,

> > To me, it's quite dissapointing that this problem even exists and from the
> > multitude of results you get out of a google search, it's been around for ages
> > and hasn't really been considered something to permanently fix. I remember
> > reading on one of the google results that one guy from Red Hat commented that
> > "not many people are affected by this so it's not worth worrying about".
> 
> Hmm .. Who is going to fix it ? RH has no interest in fixing XFS
> filesystem module .. and SGI is aware of the problem since a while 
> ... but still not fixing it.
> 
> It could be that future kernels will all have ONLY 4K stacks ... so 
> my guess is .. until SGI fixes it .. the problem will be there:
> 
> http://lwn.net/Articles/84583/
> http://lwn.net/Articles/160138/ (-> possible timeframe is 2.6.16)

Very good articles those, and I'm definately all for getting Linux working
even better.

> jarek
> 
> PS: Speaking about SuSE .. I believe you'll find other
>     things which may not be up to your taste (..as with
>     any other distro ...) : I'm personally very much in
>     favour of using RedHat/Fedora -> because of many
>     third-party compatible software repositories ... among
>     other things..

Yes, you're definately right there, I use mostly ATrpms, Dag's repo and Dries.
But I have heard that SUSE has about 60,000 packages they make available for
the dist, so the requirement for 3rd party repo's is quite diminished.

I just finished migrating one of the clusters from XFS to ext3. It took about
2 hours to do (the other cluster is taking a little longer). Surprisingly
though, using the stock RH kernel-2.6.9-22.0.1.EL has sped up the startup
sequence of the cluster (by about 20 seconds - used the "time" command before
and after to compare), which is significant considering when a failover
occurs, that 20 seconds matters. Not sure what the sl-contrib guys did from
the XFS kernel versions compared to the stock RH kernel version which would
have slowed it down by that much.

But it seems this XFS 4k issue is a kernel issue instead of an actual
distribution issue, so I can't entirely blame RH for their inaction. Although
it would be nice for them to support it.

Michael.