Network 0.0.0.0 under BGP process --- Does this change route selection criteria ?

[Sandeep Verma]

Hi Friends,

Need some of your thoughts regarding this scenario.

I have a WAN1 CE router running BGP as exterior protocol and ospf as internal (With core switch). I have redistributed the static default route into core switch  ospf process and the WAN1 router also learns it . Now, i want to advertise this default route into BGP Routing table so that other sites can learn it and i use network 0.0.0.0 for this purpose under BGP process. Apart from this router i also have the default route advertised from another site WAN2 which is also connected via MPLS(but with higher AS-Path). 

Now what happens is my ospf peering breaks between the WAN1 Router and Core Switch--- As a result default route from ospf goes missing hence the router is not advertising the default route now. This router WAN1 now has the default route from BGP neighbor which is advertised from Other Router WAN2. Hence this route gets installed in the routing table. 

This WAN1 therefore will not advertise the default route now to same neighbor from which it is learning. ---> hope this is correct ?

Now the ospf peering comes up and the WAN1 router starts learning the default route from ospf again. It however does not install this in routing table(due to eBGP having better AD than ospf) and therefore can not advertise it into the BGP routing table. 

So my question  here is why this route (default from ospf) was preferred earlier and rejected now ?-- now it is due to AD , as i understand

Does it depend on the sequence on how i configured it ... because i had this advertisement of default route from WAN1 earlier and i configured the advertisement of default route from WAN 2 recently. ...Or did the BGP on WAN1 simply reject the default route since it already had it from ospf and was advertising it...and later when ospf route got lost it simply accepted it--- as network 0.0.0.0 was not playing any role now. 

As a remedy i am planning to modify the AD for this default route from BGP neighbor to some value more than 110 . so ospf is preferred whenever the route is present and BGP preepmts this route once it starts learning back from ospf.

Thanks and Regards

Sandeep

[Robert Hillcoat]

Hi Sandeep, please upload some sort of topology drawing. 

[Robert Hillcoat · Accepted solution]

Hi Sandeep, I wouldn't change the AD of the BGP routes.

Your correct in stating that the BGP AD of 20 is preferred over OSPF of 110, therefore the BGP route remains in the routing table. 

What I would do in this situation is change the AD of the OSPF route.

Whats happening when you redistribute the default route from your core to WAN1, is a Type5 LSA is sent to WAN1, you can manipulate those type of routes by using the command "distance ospf external 19". Use this command on WAN1 router ospf process and you will achieve your goal. 

Make sure you test this in a lab environment before deployment on a live network. 

Hope this helps. 

[Sandeep Verma]

Hi Robert,

Thanks for your response!

One thing, i have the normal BGP routes redistributed on WAN1 into ospf, which  are learnt on Core Switch as LSA Type5. Would it cause any issues for these redistributed routes ?

Otherwise this seems a perfect solution.

Thanks and Regards

Sandeep

[Robert Hillcoat]

Sandeep, this command only applies to the router on which its applied.

Routes redistributed from WAN1 to Core will still appear in the Core switch routing table as E2 routes with AD of 110. 

As I mentioned before make sure you lab this up and test it before deployment. 

[Sandeep Verma]

Hi Robert,

Thanks for your response... please find attached.

Regards

Sandeep