[quagga-users 14506] Re: Advertising Default Route to particular Area using OSPF

Hello Joshua,




I didnt


Regards,

Nico




On 11/01/2016 07:41 PM, Joshua Hurst wrote:
> Hello,
>
> I am fairly new to Quagga and OSPF. I just started a new job and am 
> trying to learn as quickly as possible. I think I understand all of 
> the basic stuff, but I am having trouble understanding how to have 2 
> of my routers advertise default routes to a particular area.
>
> Here is our basic setup.
>
> Inline image 1
>
>
> We have two debian firewall servers which all of our ISP's are 
> connected to. I will refer to them as R1 and R2. They are both in Area 0
>
> R1 has this configuration for OSPF:
>  ospf router-id 10.147.1.10
>  max-metric router-lsa on-startup 30
>  max-metric router-lsa on-shutdown 30
>  network 10.3.251.0/29 <http://10.3.251.0/29> area 10.18.0.0
>  network 10.147.1.4/30 <http://10.147.1.4/30> area 0.0.0.0
>  network 10.147.1.8/30 <http://10.147.1.8/30> area 0.0.0.0
>  network 10.147.1.48/30 <http://10.147.1.48/30> area 0.0.0.0
>  area 10.18.0.0 range 10.18.0.0/16 <http://10.18.0.0/16>
>
> R2 has this configuration for OSPF:
>  ospf router-id 10.147.1.22
>  max-metric router-lsa on-startup 30
>  max-metric router-lsa on-shutdown 30
>  network 10.1.254.0/29 <http://10.1.254.0/29> area 10.18.0.0
>  network 10.147.1.17/30 <http://10.147.1.17/30> area 0.0.0.0
>  network 10.147.1.20/30 <http://10.147.1.20/30> area 0.0.0.0
>  network 10.147.1.52/30 <http://10.147.1.52/30> area 0.0.0.0
>  area 10.18.0.0 range 10.18.0.0/16 <http://10.18.0.0/16>
>
> We also have two debian firewall servers (RR1 and RR2 both in Area 0) 
> outside our local network that have openvpn connections to R1 and R2.
>
> RR1 has this configuration for OSPF
>  ospf router-id 10.147.1.9
>  max-metric router-lsa on-startup 30
>  max-metric router-lsa on-shutdown 30
>  redistribute static
>  network 10.84.0.0/22 <http://10.84.0.0/22> area 10.88.0.69
>  network 10.88.1.0/24 <http://10.88.1.0/24> area 10.88.0.69
>  network 10.88.250.0/30 <http://10.88.250.0/30> area 0.0.0.0
>  network 10.147.1.2/30 <http://10.147.1.2/30> area 0.0.0.0
>  network 10.147.1.8/30 <http://10.147.1.8/30> area 0.0.0.0
>
> RR2 has this configuration for OSPF:
>  ospf router-id 10.147.1.21
>  max-metric router-lsa on-startup 30
>  max-metric router-lsa on-shutdown 30
>  network 10.85.0.0/22 <http://10.85.0.0/22> area 10.88.0.69
>  network 10.88.1.0/24 <http://10.88.1.0/24> area 10.88.0.69
>  network 10.88.250.0/30 <http://10.88.250.0/30> area 0.0.0.0
>  network 10.147.1.12/30 <http://10.147.1.12/30> area 0.0.0.0
>  network 10.147.1.20/30 <http://10.147.1.20/30> area 0.0.0.0
>
> We have two Dell N3024 switches behind R1 and R2 that are our core 
> switches (S1 and S2). They are in area 10.18.0.0
>
> I am trying to get this set up properly for High Availability. How do 
> I have R1 and R2 advertise themselves as a default route to S1 and S2 
> without also advertising to RR1 and RR2? Can anyone point me in the 
> right direction?
So far I can tell is that it's not possible to default-originate the 
default route in OSPF on specific areas

It can be done globally with the config default-information originate 
under router ospf section ( but that's not answering your question :) )


AFAIK, OSPF is not really meant for that. I would use BGP to have 
maximum flexibility with routes.

I would suggest:

Use OSPF as what it is - IGP to "build" the map of your network -  and 
use BGP for "customer/service" traffic with route-map to announce the 
default-route ( with next-hop self maybe  ) from R1/R2 to the S1/S2 - 
and S1/S2 to accept only default routes from upstream.

That would allow you to avoid having the default-route pointing to 
RR1/RR2 as well.


> Also, is there any way to specify which one should take precedence 
> when both routes are available?
Again, I don't how to do it "clean" with OSPF - you can use different 
OSPF cost on both S1/S2 to R1/R2 links.
lets say:
S1 -> R1 cost 10
S1 -> R2 cost 100
S2 -> R1 cost 100
S2 -> R2 cost 10

Both routes are available, but as one next-hop will have a higher cost - 
it will break ECMP ( but not the redundancy )

A cleaner way can be done again with BGP + route-map ( that can set bgp 
attributes like MED / WEIGHT ) based on community for example at the 
R1/R2 or S1/S2 level.
Some people would prefer to fix route choice at the R1/R2 level, others 
at the S1/S2 level - depends on the habit I guess.

Regards,
Nicolas

Ps: what are the default-routes for R1/R2 ? one or more ISP ? what 
happens if all uptreams goes done on R1 or R2 ?


> We have two other debian firewalls connected through openvpn that are 
> set up similarly to RR1 and RR2. We cannot afford to have ALL the 
> traffic from the other locations coming through our main location.
>
>
> _______________________________________________
> Quagga-users mailing list
> Quagga-users-UOy77sIEA+cAd7ICUelF/[email protected]
> https://lists.quagga.net/mailman/listinfo/quagga-users

_______________________________________________
Quagga-users mailing list
Quagga-users-UOy77sIEA+cAd7ICUelF/[email protected]
https://lists.quagga.net/mailman/listinfo/quagga-users