Hi Claudio,

On Fri, 24.08.2007 at 11:05:04 +0200, Claudio Jeker <cjeker@diehard.n-r-g.com> wrote:
quoted text
> All redistributed networks are originated with a nexthop of the ospf
> router. That is how it works for the moment and changing that is not
> simple as you need to know if the nexthop you intend to use is reachable
> by the other ospf routers. Make sure you only announce static routes that
> are behind the router that redistributes them.

thank you very much for the explanation. This is then an
"implementation weakness", and it seems to preclude the (intended) use
for the following scenario:


 * router A announcing a route via router B (which is not an OSPF
   speaker)


I'm not quite sure if your statement also goes for "connected"
networks. I'd like to be able to announce networks dynamically if they
"go up", like this:


 network N1 --- C ---------+------s1-----A----+---------- network N2
                           |                  |
			   |                  |
			   +------s2-----B----+
			   |                  |
			   |                  |
			   +--E---s3-----D----+


I'm in N2, reachability of N1 is what bothers me. Bandwidth (for OSPF
weight calculations) is like this: s1 = s2 >> s3.

A and B have a (static?) route to N1 via C, and if s1 or s2 are down, A
or B should stop announcing their route to C, respectively. So far, I
have made C announcing N1 to A or B (currently, only one of them is
active at any one time). The route via D (OSPF speaker) and E (dumb) is
also mostly available, but this is the least preferred route, as s3 is
only a thin backup line...  nevertheless, sometimes the route via D is
the one injected into the routing table, while fast s1 or s2 are
ignored.


I'll probably start using snapshots really soon now. ;-}



Best,
--Toni++