OSPF routing between HUB & Spokes

[karamalomari]

I have the attached setup and the issue is that when a spoke tries to communicate with another spoke the communication is dropped.

The question is that when a spoke tried to communicate with another spoke, can we force the traffic from 1 spoke to another spoke go through 1 of the main HUB's. The connection that we have is point-to-multipoint and there are no direct connection between the spokes.

example: Spoke-1 --> HUB --> Spoke-2 and vise versa.

Attached is the layout along with configuration.

Please let me know how to force the traffic from all the Spokes to go 1 of the HUB's and then to another Spoke rather than trying to directly spoke --> spoke without going through the HUB's.

[MHM Cisco World]

I don't see the config of tunnel in Spoke ? 

[MHM Cisco World]

anyway until you share config 

you must sure that you see prefix of other spoke in spoke-A (example) next-hop is Hub interface, also you must sure that no phase2 & 3 of DMVPN is run
all above make traffic always between Spoke pass though Hub.

[Giuseppe Larosa]

Hello @karamalomari ,

>> The connection that we have is point-to-multipoint and there are no direct connection between the spokes.

if you have an E-TREE connectivity model

you need to use

ip ospf network point-to-multipoint

and

ip ospf cost at interface level can give you a way to have a primary path via links to Hub1 for example

with default configuration the OSPF type is broadcast and the spokes attempt to partecipate in DR/BDR election

you can check this using

show ip ospf interfaces on both spokes and hubs

you need also to adjust OSPF auto-refeference bandwidth or all links will have cost 1

router ospf 1

auto-reference bandwidth 1000000

You need to study OSPF in more depth.

Hope to help

Giuseppe

[Georg Pauwen]

Hello,

you could use policy routing, this would override any OSPF defined routing. The configs would be:

Spoke-01

ip access-list extended Spoke-01_Spoke-02
permit ip 172.50.10.0 0.0.0.255 172.20.10.0 0.0.0.255
permit ip 172.50.10.0 0.0.0.255 172.20.11.0 0.0.0.255
permit ip 172.50.11.0 0.0.0.255 172.20.10.0 0.0.0.255
permit ip 172.50.11.0 0.0.0.255 172.20.11.0 0.0.0.255
!
route-map PBR permit 10
match ip address Spoke-01_Spoke-02
set ip next-hop 172.16.150.1
!
interfaceGigabitEthernet0/1
policy route-map PBR
!
interfaceGigabitEthernet0/2
policy route-map PBR

Spoke-02

ip access-list extended Spoke-02_Spoke-01
permit ip 172.20.10.0 0.0.0.255 172.50.10.0 0.0.0.255
permit ip 172.20.10.0 0.0.0.255 172.50.11.0 0.0.0.255
permit ip 172.20.11.0 0.0.0.255 172.50.10.0 0.0.0.255
permit ip 172.20.11.0 0.0.0.255 172.50.11.0 0.0.0.255
!
route-map PBR permit 10
match ip address Spoke-02_Spoke-01
set ip next-hop 172.16.130.1
!
interfaceGigabitEthernet0/1
policy route-map PBR
!
interfaceGigabitEthernet0/2
policy route-map PBR

[paul driver]

Hello
Your current topology isn’t optimal as the hubs won’t have any ospf adjacency to each other, however to establish communication between spoke via the hubs via ospf you need to change the network type to a network point-to-multipoint non-broadcast (NBMA) this way there is no automatic neighbor creation and you will need to specify a specific neighbor for adjacency

Lasty append a high ospf interface costing on the spoke rtrs to the less prefferd interface.

Example:
Hubs
int x/x
ip ospf network point-to-multipoint non-broadcast

router ospf  xx
neighbor <spoke 1 ip >
neighbor <spoke 2 ip >

Spokes
int x/x
ip ospf network point-to-multipoint non-broadcast

int x/y  < less preffered path>
ip ospf network point-to-multipoint non-broadcast
ip ospf cost 1000