Hi Robert and Roger,

First, thanks for your responses. I just wanted to follow-up and let you know how things worked out.

In retrospect, I was making things too complicated. I should have RTFM (the installation instructions), which I had not referred to for quite some time. So the easiest thing to do to temporarily add nodes is to just use etherboot on a floppy disk! That way you do not have to go in and mess with bios setting for PXE for what is only supposed to be a temporary node. The ROM-o-matic (as pointed out in the installation instructions) is really great:
It took a couple of tries to get the right combination of settings, but it is so easy to do with this service. A couple of things that I discovered though:
  1. if you want the node to immediately boot from the diskette, that is without asking for input, then ASK_BOOT should be unset (blank)
  2. if the node has multiple NICS, then you will need to set the BOOT_INDEX to get the correct NIC (assuming that only one is connected to the network, as in the case where on NIC is 100/1000 and the other is only 100).
So I was easily able to add two additional nodes to my existing 3 node cluster. Obviously I had to point fstab to the right swap partitions to get their swap spaces. So now I have 10 cpus working on an few animation sequences.


-Jack de Valpine

Robert Wachinger wrote:
On Wed, Mar 22, 2006 at 11:33:00PM +1100, Roger Tsang wrote:
You can network boot these machines into the OpenSSI cluster and use
the existing swap partitions on these temporary nodes but not mount
the local filesystems.

Is this so simple?
As far as I know, you have to run ssi-addnode to add some node to the
cluster (you need among others entries in /etc/clustertab,
dhcpd-configuration and a newly built initrd, which is aware of the new

And why should it not be possible to mount the local filesystems?


On 3/22/06, Jack de Valpine <jedev@visarc.com> wrote:

I have a 3 node SSI cluster running currently, with two of the nodes
using PXE to boot. I would like to figure out how to add some additional
nodes to the cluster on a temporary basis. These additional nodes
already have linux installations on them. Ideally, I would like to
figure out if there is a way to reboot some of these additional nodes so
that they will join cluster for that reboot session (eg a further reboot
would bring a node back up in its original configuration). Is there a
way to pass some temporary instructions via grub? Can grub be used to
select a pxe boot session

In case, that your BIOS does not support PXE-Boot, you could get a
.zlilo file from http://rom-o-matic.net/5.4.1/ for your network card,
and add something like
title           Etherboot
root            (hd0,0)
kernel          /eb-5.4.1-rtl8139.zlilo
into menu.lst

Works fine ;-)
I entered such nodes during ssi-addnode as type "etherboot".

                             Also can I use some of the existing
partitions on the temporary nodes such as swap partitions? I think that

Just enter your partitions into /etc/fstab accordingly.

the nodes I would be adding are running FC3 (that is the
filesystem/partitions have been built with a more current FC version
than the SSI cluster with I believe is based on FC2).

Should there be some incompatibilities?

Regards, Robert

This SF.Net email is sponsored by xPML, a groundbreaking scripting language
that extends applications into web and mobile media. Attend the live webcast
and join the prime developer group breaking into this new coding territory!
Ssic-linux-users mailing list


# Jack de Valpine
# president
# visarc incorporated
# http://www.visarc.com
# channeling technology for superior design and construction