Subject: Re: should the default route get a new interface
To: NetBSD Networking Technical Discussion List <tech-net@NetBSD.ORG>
From: Markus A. Boeing <email@example.com>
Date: 07/28/2001 20:53:06
I wasn't following the whole track, so I may be totally of topic. If so,
From reading you mail I do not see a requirement for routing or a default
route. If you just have multiple network cards in your box and a single IP
network/sub-network per NIC, then all networks are connected networks. In
an IP network there is no need for routing protocols or a default route to
communicate with connected networks. If you have IP networks/sub-networks
that are not directly connected to your box then you need either a default
route or a routing protocol to reach them. In that case dynamic routing
protocols -such as OSPF or RIPv2- can provide significant benefit.
My question would be, why do you need a default route if all networks are
directly connected networks?
At 02:35 28.07.2001, Greg A. Woods wrote:
>[ On Saturday, July 28, 2001 at 09:56:27 (+0900), firstname.lastname@example.org wrote: ]
> > Subject: Re: should the default route get a new interface automatically?
> > amount of dynamism does not have anything to do with justification
> > for routing daemons. all you need to do is
> > - advertise routes (like just default) from two of your routers
> > - listen to them on your server (routed -q)
> > then
>But I have no routers in this case -- just multiple networks. No
>routing happens, the server just sits on multiple networks (production,
>test, etc.). There are no routes to advertise in the first place.
>(in fact I want to do even more of this -- I've an entire separate
>DEChub900 with switches and hubs and stuff that I want to set up for
>testing equipment and systems in a mostly isolated environment; a safe
>place where I can let wild things loose, so to speak)
>None of this changes the fact that the default route learns the
>interface dynamically now, but then gets it stuck. One shouldn't need a
>routing daemon to unstick it, especially not when the kernel can
>easily dynamically re-learn similar interface flags on other routes.
>(I won't even mention similar appletalk problems yet... :-)
> Greg A. Woods
>+1 416 218-0098 VE3TCP <email@example.com> <firstname.lastname@example.org>
>Planix, Inc. <email@example.com>; Secrets of the Weird <firstname.lastname@example.org>