It sounds like you might be doing something like PelicanHPC, which I have
posted about on this list in the past. I find that exporting a squashed fs
works pretty well. Klaus Knopper backed up this empirical observation with
an argument, a while ago. I think that it had to do with properly balancing
CPU demands on the nodes of the cluster with network traffic, but the
details escape me now. You might want to search the Knoppix mail list for
On Feb 5, 2008 3:44 PM, Chris Fanning <christopher.fanning at gmail.com> wrote:
> I'm new on the list but I have had some success using casper on etch.
> We would like your opinion on a cluster setup that we're think of.
> The idea is to boot all diskless servers over the network using a
> single filesystem image.
> Does that sound like a good idea?
> We have managed to do it in the workshop using a squashed filesystem,
> but we would like to avoid the overhead and use instead a non-squashed
> I've been trying to do it, but I'm not having much luck.
> How is it done?
> debian-live-devel mailing list
> debian-live-devel at lists.alioth.debian.org
-------------- next part --------------
An HTML attachment was scrubbed...