Re: [OOPS] 2.6.9-rc4, dual Opteron, NUMA, 8GB

[Jeff Garzik]

On Wed, Oct 13, 2004 at 01:49:08PM -0700, Brad Fitzpatrick wrote:
> On Wed, 13 Oct 2004, Jeff Garzik wrote:
> 
> > Brad Fitzpatrick wrote:
> > > I was just about to mail and report that disabling NUMA does help:
> > >
> > >    NUMA + mke2fs on LVM:      OOPS  (mailed earlier)
> > > no NUMA + mke2fs on LVM:      okay
> > >    NUMA + mke2fs on sdb1:     OOPS  (below)
> > > no NUMA + mke2fs on sdb1:     okay
> > >
> > > no NUMA + mount e2fs on LVM:  okay
> > > no NUMA + mount e2fs on sb1:  okay
> > >    NUMA + mount e2fs on LVM:  okay
> > >    NUMA + mount e2fs on sb1:  untested, assume okay
> >
> >
> > Honestly it smells like a devmapper bug, with the pertinent test case
> > being some sort of behavior not normally generated by filesystem use
> > (the most common use case for block devices).
> 
> Except for the data point above:
> 
> NUMA + mke2fs on sdb1:     OOPS  (below)
> 
> That's with devmapper totally out of the picture.

Ahh, indeed.

I think Andi did NUMA on x86-64?

	Jeff



-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/