[Xorp-users] Setting default routes

[Pavlin Radoslavov]

> atanu@icsi.berkeley.edu said:
> > I tried the following in my config file, it worked for me.
> > ---------------------------------------- protocols {
> >   static {
> >      route4 0.0.0.0/0 {
> >          nexthop:  144.124.0.2
> >          metric: 1
> >      }
> >   } } ---------------------------------------- 
> 
> That lookss exactly the same as what I believe I had
> added yesterday when I could not seem to get it to work.  I'll
> try again.   Slight complication in so much as we subnet
> 144.124.x.x (at 22 bits) and so the interface to the default
> is on a subnet of this, but I can't see that should change things.
> 
> I have NOT enabled the unicast FEA as I don't actually want this box
> to unicast forward between its ports, I plan to just get it
> to act as a multicast router.  I wonder if that interacts.
> Anyway, I'll try to static default route again.

Dave,

If you are going to use the router only for IPv4 multicast routing,
then you don't have to enable the unicast forwarding.

However, I should let you know that if you are going to use it for
IPv6 multicast routing as well, and if you are using a KAME-based
stack (e.g., *BSD), then the semantic is that you must have the
following flag set:
  sysctl net.inet6.ip6.forwarding=1

The above flag enables both IPv6 unicast and multicast forwarding.
You can set it from the command line (as a root) before starting
XORP, or from the XORP configuration file:

fea {
    /* enable-unicast-forwarding4: true */
    enable-unicast-forwarding6: true
}

Hence, because of the difference in the semantics between IPv4 and
IPv6, in general I recommend that both IPv4 (net.inet.ip.forwarding)
and IPv6 (net.inet6.ip6.forwarding) forwarding flags are enabled.
In your case, if none of your unicast routers have your multicast
router as a next-hop router for a destination, then enabling the
IPv4 unicast forwarding flag shoudn't make any difference.

Pavlin

> atanu@icsi.berkeley.edu said:
> > I believe that we only speak RIPv2.
> 
> I had come to that conclusion, the docs implied that and
> I could see multicast RIPv2 requests coming out of the xorp
> box when I did enable RIP.  It was ignoring
> RIPv1 announcements that were arriving.
> 
> 
> atanu@icsi.berkeley.edu said:
> > The simplest way of discovering all the settings for a protocol is to
> > use the xorpsh. In configure mode it will show all the settings along
> > with help text.
> 
> That lets you discover their existence, but it does not give
> precise descriptions of their role etc.  Some have names
> which might be interpreted in more than one way I felt.
> 
> 
> atanu@icsi.berkeley.edu said:
> > We are intending to make some sample configuration files available.
> > Your configuration file would be a great candidate.
> 
> assuming I can get it to work :-)
> 
> I'll go and have another play....
> 
> Dave Price