[quagga-dev 4130] Re: Inconsistent "inactive" route in OSPF

Erik Tromp erik_tromp at hotmail.com
Mon May 8 18:58:13 BST 2006


Okay you have a point...

So now I go to the guys at OLSR and tell them they made a
piece of #$%^     

;-)

> Did you wait for > 1 minute after step 6?
Nope. So the route would have become inactive after 1 minute?

>    C>* 10.0.3.0/24 is directly connected, eth1
>    K>* 10.0.3.3/32 via 10.0.3.1
Indeed this will also be the case often. Multi-hop connections do occur
in the wireless world. So will you support it :-) ? Then I have a case
to get OLSR patched (or do it myself) to not enter pointless routes.
Then, there will be interopability! 

Anyway thanks for your patch. I will try it tomorrow. Which version
can I patch it on?

Erik

----- Original Message ----- 
From: "Paul Jakma" <paul at clubi.ie>
To: "Erik Tromp" <erik_tromp at hotmail.com>
Cc: "Quagga Dev" <quagga-dev at lists.quagga.net>
Sent: Monday, May 08, 2006 6:22 PM
Subject: [quagga-dev 4126] Re: Inconsistent "inactive" route in OSPF


> On Mon, 8 May 2006, Erik Tromp wrote:
> 
> > I don't understand. All i am saying is: look:
> >
> > After step 6) we have:
> >  C>* 10.0.3.0/24 is directly connected, eth1
> >  K>* 10.0.3.3/32 is directly connected, eth1
> >  S>* 10.0.43.0/24 [1/0] via 10.0.3.3, eth1
> >
> > After step 10) we have
> >  C>* 10.0.3.0/24 is directly connected, eth1
> >  K>* 10.0.3.3/32 is directly connected, eth1
> >  S   10.0.43.0/24 [1/0] via 10.0.3.3 inactive
> >
> > What's the difference??
> 
> Did you wait for > 1 minute after step 6? Likely the periodic 
> next-hop scanner would come along and turn 6 into 10 (that's what 
> /should/ happen at least).
> 
> > Talking about silly things. Would you beleive there are certain area
> > in the world where people wear clothes? Really :))
> 
> Well, it /is/ a pointless route. The prefix falls within the 
> connected subnet and has the interface as target, there really is 
> /no/ point adding it AFAICT.
> 
> > No kidding, the whole point of the /other/ routing protocol is to 
> > generate these silly on-link host-only routes. In that way the 
> > hidden node problem of wireless IP networks can be tackled. In a 
> > wireless network, all hosts may be within the same IP-subnet, but 
> > that does not mean that each host can reach every other host via 
> > one hop, "on-link". OLSR detects such situations and generates 
> > multi-hop routes within the same IP subnet. Nifty, ehh?
> 
> It really /need/ only install the host-routes when the address 
> concerned is on a 'non-native' interface. I.e. there is 0 point to:
> 
>    C>* 10.0.3.0/24 is directly connected, eth1
>    K>* 10.0.3.3/32 is directly connected, eth1
> 
> If 10.0.3.3 really were hidden, and you had it routed via another 
> /host/:
> 
>    C>* 10.0.3.0/24 is directly connected, eth1
>    K>* 10.0.3.3/32 via 10.0.3.1
> 
> *then* there'd be a point, and that case I /would/ like to support. 
> But that's different from the above case. ;) For which I /think/ the 
> patch I gave you should work.
> 
> What to do about the second case, I don't know yet - would like to 
> see it first :).
> 
> regards,
> -- 
> Paul Jakma paul at clubi.ie paul at jakma.org Key ID: 64A2FF6A
> Fortune:
> vi needs to be upgraded to vii
> _______________________________________________
> Quagga-dev mailing list
> Quagga-dev at lists.quagga.net
> http://lists.quagga.net/mailman/listinfo/quagga-dev
> 
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: Why-is-this-spam.txt
URL: <http://lists.quagga.net/pipermail/quagga-dev/attachments/20060508/0b57eb69/attachment-0001.txt>


More information about the Quagga-dev mailing list