--
------------------------------------------------------
Brian Elliott Finley Mobile: 630.631.6621
gpg --keyserver wwwkeys.pgp.net --recv-keys 10F8EE52
------------------------------------------------------
--- Begin Message ---
> Let me say thanks for your well thought out email.
You're welcome.
> Also, do you know Jan Lindheim and/or Mark Bartelt?
Only by name and an email or two.
>
> Comments below...
>
>
> David Mathog wrote:
>
> >Greetings,
> >
> >I use systemimager and boel to replicate nodes on our beowulf.
> >In order to speed things up I first modified boel to include
> >"dolly", which is a daisy chain copier,
> >
> Cool. I've heard of "dolly", but have never played with it.
>
> >instead of rdist, which
> >isn't.
> >
> I presume you mean rsync?
Yes, sorry.
> But your point is a good one -- rsync does
> not use a daisy chain.
>
> >That helped considerably but dolly had a few problems
> >which I've hopefully remedied in the much simplified "nettee",
> >which is derived from dolly.
> >
> * Is nettee software that you wrote?
Yes
> * Based on the code from dolly?
Yes
> * Is it Open Source?
Yes, GPL 2, see here:
http://saf.bio.caltech.edu/nettee.html
It's up to 0.1.4 at the moment.
> >
> >Following the instructions in your boel article at linuxdevices.com
> >and the man pages this was added to dhcpd.conf on the dhcp server:
> >
> >option nextnode code 200 = text;
> >
> >
> Hmmm. This will work if you're imaging all nodes at once, but what
> happens if you're only imaging a portion of the nodes, and they don't
> happen to be adjascent to each other according to their dhcp info?
> (other than re-writing the dhcpd.conf file)
>
There is another way, I was essentially just asked that by somebody
else and I'm going to forward my response to his question to you
next.
> >
> >In other words, the nextnode line in the dhclient.conf is required
> >to retrieve the 200 value, but it is stamped with "unknown-200"
> >instead of "nextnode".
> >
> >
> I think there is a way to handle this in dhcp3 by defining your own
> option types. But option-200 works just fine too. ;-)
That's eventually what I used.
> >
>
> I'd be interested in considering this for inclusion in the upstream
> source, as an alternative transport. Especially if there is a good
> solution to the "I have no adjascent neighbor" question.
Fine with me. There is a solution to the no adjascent neighbor
but I've not yet tested it myself. It's coming in the next message.
Regards,
David Mathog
[EMAIL PROTECTED]
Manager, Sequence Analysis Facility, Biology Division, Caltech
--- End Message ---
