Subject: Re: Posible virc(8) implementation
To: NetBSD-current Discussion List <current-users@netbsd.org>
From: Greg A. Woods <woods@weird.com>
List: current-users
Date: 05/04/2000 18:56:11
[ On Thursday, May 4, 2000 at 11:33:30 (-0700), Jonathan Stone wrote: ]
> Subject: Re: Posible virc(8) implementation 
>
> What I have been doing since rc.conf was created is to keep all
> host-identifying info (basically /etc/{ifconfig.*,myname,mygate} in
> the "old" files, and _everything_ else in /etc/rc.conf.
> 
> I've found this a very valuable way to "clone" specific machines, or
> populate a farm of "identical" machines.  I seem to do that a lot,
> both in an academica lab environment and in a startup.
> z
> I really, really, dont want to lose the abillity to centralize
> information that way. virc is *not* a substitute for that; the people
> who've claimed that it is are just asking to be flamed.

I don't see the problem.  You still only have to edit exactly the same
number of files to make your cloned machine unique.  The cloning process
itself has already copied all of the other files for you -- what
difference does it make to you whether during cloning you have copied
one or a dozen or more "common" configuration files?

If you're talking about re-cloning the machines to propogate a
configuration change across all related machines then it should simply
be a matter of creating a manifest of common files and using a simple
one-line script to always copy all of the common files with every
update.  Rdist would seem to be your best friend here, though of course
any such tool will force you to be a little less ad hoc in your
configuration management (which is supposed to be a *good* thing!).

-- 
							Greg A. Woods

+1 416 218-0098      VE3TCP      <gwoods@acm.org>      <robohack!woods>
Planix, Inc. <woods@planix.com>; Secrets of the Weird <woods@weird.com>