Re: swapper: page allocation failure. - random reboot problem

From: Haar János
Date: Mon May 22 2006 - 17:06:42 EST



----- Original Message -----
From: "Con Kolivas" <kernel@xxxxxxxxxxx>
To: "Haar János" <djani22@xxxxxxxxxxxx>
Cc: <nickpiggin@xxxxxxxxxxxx>; <cw@xxxxxxxx>; <linux-kernel@xxxxxxxxxxxxxxx>
Sent: Monday, May 22, 2006 5:12 PM
Subject: Re: swapper: page allocation failure.


> On Tuesday 23 May 2006 01:08, Haar János wrote:
> > ----- Original Message -----
> > From: "Con Kolivas" <kernel@xxxxxxxxxxx>
> > > Try with one of the alternative vmsplit options that gives you more
> >
> > lowmem?
> >
> > > That might break certain applications though.
> >
> > total used free shared buffers
cached
> > Mem: 4049724 4021196 28528 0 16384
3217288
> > Low: 4049724 4021196 28528
> > High: 0 0 0
> > -/+ buffers/cache: 787524 3262200
> > Swap: 0 0 0
> >
> > This is an 64 bit machine, the "concentrator".
> >
> > It looks like use all, the 4G ram as "lowmem".
> > If i replace the cpu on my nodes to 64bit capable ones, i can use all
the
> > memory as buffer-cache? :-)
>
> Heh yes indeed. It's only if you're stuck on 32bit for whatever reason
that
> you'd need a different vmsplit. There is no need for highmem when 64bit
> allows bazillions of bytes of lowmem :)

OK, it is enough, to switch to 64bit, thanks!

But i have a little problem.
My node #3 reboots again.

At this point i have run out of ideas. :-(

This is checked already:

- the complete hardware, except the 12 hdd. (smart reports, no errors at
all, 4x ide + 8xSATA all 300GB.)
- the SMP race. (checked with non-smp kernel)
- APIC/ACPI (tested with non... kernel)
- the e1000 driver (tested with realtek gige adapter)
- the complete filesystem, OS (NFS-ROOT, and copy between nodes.)
- the memory allocation proble, (checked with debug-kernel, and rised
min_free_kbytes)

The systems only service is nbd. (nbd-server serving md0, raid4 array)

Anybody have an idea?

Please let me know!

Thanks,
Janos



>
> --
> -ck
> -
> 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/

-
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/