[Techtalk] Re: kernel problems on a cluster

[MARAPAO,CHRISTINE JEREME DEBELEN]

Hello Maria,

Have you tried posting at the beowulf cluster mailing list?
Their website's at: www.beowulf.org. You might get help there as well.

Tintin

> Hi there,
>
> I have a cluster of diskless machines, that boot off the network.  I
> have been upgrading them slowly.  I have two machines that are running
> successfully with a new motherboard and processor.  I have been trying
> to install the exact same model motherboard and processor (and at least
> once exactly the same motherboard and processor) into *any* of the
> other
> machines, and I can't boot with it.  The rest of the equipment,
> ethernet
> cards, video cards are exactly the same, as is the file system, and all
> of the machines will work with the old motherboard.  I have literally
> switched out all of the stuff in the boxes except the power supply.  It
> seems like there has to be something different about those two
> machines,
> but I can't come up with anything.  Maybe if I could figure out what
> these errors have to do with, I could figure it out, but googling gives
> me a bunch of stuff that I don't understand in the least.  Here are the
> errors I get:
>
> Unable to handle kernel paging request at virtual address f8000000
> printing eip:
> c0108cfc
>
> I found this eip in a list for slackware that says: setup_x86_irq
>
> sometimes I get this error, which seems related:
> Unexpected IRQtrap at vector 4d
>
> The only thing that is wierd about the new boards is that we are not
> using the built-in ethernet port.  I did have to do some fussing to get
> it to use the ethernet card to boot over the network, but I made the
> same changes to files of all the computers, and this seems to be
> working
> fine on the two successful machines.
>
> Anyone have any ideas?  Or can anyone tell me another good place to
> post, and what all information I need to include?
>
> thanks a bunch,
> maria
>