Node cloning

Nordwall, Douglas J Nordwall at
Wed Apr 4 13:27:24 PDT 2001

the kickstart method is the one used by the npaci rocks distribution. they put a
rather large post config section and do a lot of the major tweaking there. I've
put kup a couple of npaci rocks clusters an am quite happy with it. Indeed, I
stole several chunks of their script for other projects I work on at the lab.

> -----Original Message-----
> From: Robert G. Brown [mailto:rgb at]
> Sent: Wednesday, April 04, 2001 1:10 PM
> To: Richard C Ferri
> Cc: Luca Frediani; beowulf at
> Subject: Re: Node cloning
> On Wed, 4 Apr 2001, Richard C Ferri wrote:
> >
> > Luca,
> >      There are two good solutions that I am aware of to 
> clone nodes in a
> > beowulf cluster.   If all your nodes are basically 
> identical (except for IP
> > information, and the size of the harddrive) you should look at
> > Systemimager, an open source project sponsored by VA Linux:
> >           If 
> your nodes
> > differ significantly, look at LUI, an open source project 
> sponsored by IBM,
> > at        There are mailing lists
> > associated with both projects  to help you in your time of 
> need... Rich
> I'd add two more ways.  One is to use Scyld.  One doesn't 
> exactly clone
> nodes, but what you end up with is functionally the same and even
> simpler.  The second is to just use Red Hat and kickstart.  A 
> kickstart
> script can be written for a generic node, and a slightly hacked boot
> floppy built (basically modified to go into kickstart by 
> default after a
> timeout instead of any of the interactive startups) that will 
> install a
> node on boot.  This latter is my favorite, as it scales well, 
> gives you
> a complete choice as to which RPM's to include, and works for 
> arbitrary
> clusters and even departmental networks of moderately 
> dissimilar desktop
> hosts.  With the boot/kickstart install floppy, there is no real need
> for a node to have a head -- insert floppy, power up, wait thirty
> minutes (or less), remove floppy, reboot.  You don't even 
> have to have a
> keyboard -- the second boot can safely be a power cycle since 
> the system
> typically fsck's on the first boot anyway.
>    rgb
> -- 
> Robert G. Brown	                       
> Duke University Dept. of Physics, Box 90305
> Durham, N.C. 27708-0305
> Phone: 1-919-660-2567  Fax: 919-660-2525     email:rgb at
> _______________________________________________
> Beowulf mailing list, Beowulf at
> To change your subscription (digest mode or unsubscribe) 
> visit

More information about the Beowulf mailing list