[quagga-dev 5448] Re: [PATCH] RFC 2328, chap 8.1:

Joakim Tjernlund joakim.tjernlund at transmode.se
Tue Jun 3 16:53:19 BST 2008


On Tue, 2008-06-03 at 11:43 -0400, Andrew J. Schorr wrote:
> On Tue, Jun 03, 2008 at 05:34:42PM +0200, Joakim Tjernlund wrote:
> > No, you are making sense. I was aiming for the "single shared address"
> > but I see now that it has some drawbacks.
> > The "pair of local and remote" address does not work for me. The need
> > for specifying the remote address requires too much from the user. Also
> > what do you do when you have several remote nodes you want to connect
> > too with unnumbered i/f's? 
> 
> OK, I think this is progress, we are now understanding the problem the
> same way.  Perhaps this should work similarly to the existing
> passive-interface behavior?  So there could be a new ospf subcommand:
> 
>    [no] unnumbered-ptp {default|<interface name>}
> 
> If you want to use the unnumbered behavior on all PtP links, just specify:
> 
>    unnumbered-ptp default
> 
> Otherwise, specify each PtP interface explicitly.
> 
> Does that get the job done?

Oh yes, that would work for me. The only problem I got ATM is that
once I enable unnumbered PtP, it stops working with our old zebra based
ospf. That is a separate issue though that I need to think some more
about. Maybe we will need an extra option for this or just do a local
hack.

> 
> Regards,
> Andy
> 



More information about the Quagga-dev mailing list