Subject: Re: dhclient in rc*/netstart
To: Jason Thorpe <thorpej@nas.nasa.gov>
From: Curt Sampson <cjs@cynic.net>
List: tech-net
Date: 01/27/1999 14:32:00
On Wed, 27 Jan 1999, Jason Thorpe wrote:

>  > Now I can fix half the problem with some code I'm going to put in
>  > to /etc/rc mount all local (ffs, basically) critical filesystems
>  > before netstart is run. However, if your /var is an NFS filesystem...
> 
> DANGER WILL ROBINSON!
> 
> It's certainly possible for the following to occur:
> 
> 	mount server:/foo /foo
> 	mount /dev/sd1a /foo/bar
> 
> (and, as sick as it seems, I have even seen some reasonable uses for
> such a configuration)

Sure, but all of the following conditions have to be true for breakage to
ensue:

    1. /foo/bar is a critical filesystem (as would be /foo, obviously).

    2. /foo is not a local filesystem (network mounted)

    3. /foo/bar is a local filesystem (not network mounted)

This sort of thing is weird enough that I'd say we should leave it
to the sysadmin to rewrite rc as he needs to for his configuration.

If we do want to support this, we could do so by having two sets
of critical filesystems, one mounted before netstart, one after.
This option would also suit me fine; do others think it's worthwhile?
(That does, of course, raise the question of which set of critical
filesystems /usr and /var should be in by default.)

cjs
--
Curt Sampson  <cjs@cynic.net>   604 801 5335   De gustibus, aut bene aut nihil.
The most widely ported operating system in the world: http://www.netbsd.org