Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 8 Jul 1999 10:01:37 -0700 (PDT)
From:      "Rodney W. Grimes" <rgrimes@gndrsh.aac.dev.com>
To:        peter@netplex.com.au (Peter Wemm)
Cc:        cvs-committers@FreeBSD.org, cvs-all@FreeBSD.org
Subject:   Re: cvs commit: src/etc rc.network src/etc/defaults rc.conf
Message-ID:  <199907081701.KAA41342@gndrsh.aac.dev.com>
In-Reply-To: <19990707163445.6A7F678@overcee.netplex.com.au> from Peter Wemm at "Jul 8, 1999 00:34:45 am"

next in thread | previous in thread | raw e-mail | index | archive | help
> "Rodney W. Grimes" wrote:
> > > peter       1999/07/07 05:49:48 PDT
> > > 

24 hours have elapsed, I've seen 2 (1 of them me) requests to revert,
none in support other than you.  Please revert...

> > >   Modified files:
> > >     etc                  rc.network 
> > >     etc/defaults         rc.conf 
> > >   Log:
> > >   Do away with ${network_interfaces} in rc.conf.  Just use `ifconfig -l` to
> > >   get a list of interfaces, and then automatically configure them if
> > >   ${ifconfig_${ifn}} or /etc/start_if.${ifn} exists.
> > >   
> > >   This makes it a lot easier to deal with machines that constantly change
> > >   their network configuration as you can leave ifconfig settings for all
> > >   the possible cards - just the ones that are present will be configured.
> > >   
> > >   Revision  Changes    Path
> > >   1.48      +12 -3     src/etc/rc.network
> > >   1.16      +1 -2      src/etc/defaults/rc.conf
> > 
> > I strongly dislike this idea.  If I list an interface in network_interfaces
> > during boot:
> > 
> > a)  I want to see an error if the physical device is missing
> > 
> > b)  It might not even exist in an ifconfig -l at this time, it
> >     just might be getting created by /etc/start_if.X
> > 
> > IMHO trying to make this smart is going to lead you down the road to
> > the problems that happen on Micro$oft systems.
> 
> The original version I did merged ifconfig -l with ${network_interfaces)
> and suppressed duplicates.  That would have generated diagnostics on a
> missing interface.  I ended up leaving it out since all our existing
> interfaces appear at boot time, even if they have not been used.  It seemed
> a bit of overkill, but if people want this, I'm happy to revive that...

I don't think that would be much better.   I've shown you how to make
the current (before your commit) system do what you want, and still do
what I want.  What more do you need?

network_interfaces="`ifconfig -l`"

> 
> On that note, I think there are better ways to do dynamic interface
> configuration.  On one box, I have a start_if.ppp0 script.  The only
> problem is that there's no telling if it really is going to get assigned
> ppp0, especially if there is a second dialup done at boot.  ie: keying this
> hook by interface name for dynamic interfaces doesn't work very well except
> in trivial cases.  I would prefer to have /etc/netstart.d/* or something
> like that for dynamic interfaces that need to be started before the rest of
> the networking.  This would be more important down the track with dynamic
> tun, ppp and sl interface creation.

I'm aware of the dynamic ppp/slip problems, that is not what I was complaining
about.  It's the fact that your changes would make it very hard for me to
get the 4 or so ETInc routers I have working correctly.  There is no ethX
device until after certain commands that are run from /etc/start_if.ethX
are run.

I know I could go to starting this from /etc/rc.local, but that would be
too late for the other stuff in /etc/rc.network, like gated, ntpdate, etc.

Again, please revert... or show me how to get the functionality we already
had back without major surgery...
-- 
Rod Grimes - KD7CAX - (RWG25)                    rgrimes@gndrsh.dnsmgr.net


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe cvs-all" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199907081701.KAA41342>