How is it possible?

[ugyentashi]

R2 is the border router of a company and is running simple eBGP with the outside world. It advertises its

103.7.116.0/24

into BGP.

R2 has a static route pointing towards the firewall for this

103.7.116.0/24

R2 is connected to the firewall with a

103.7.116.128/25 infra-network

R2 inside - 103.7.116.129/25

Fw-outside - 103.7.116.144/25

By logic we know that anything encompassing the

/25

subnet of this network, R2 will learn via connected and anything outside the

/25

but within the

/24

will be known via static (pointed towards the firewall).

The question here is when an outside user tries to communicate (here its is ping) to an address of

103.7.116.143

[which resides on the firewall as a host server], R2 should be learning this via the connected

103.7.116.128/25

So the communication to the right server on the firewall should fail, for the reason that it doesn't use the correct static route pointed to the actual server on the firewall.

How is it possible?

[Harold Ritter · Accepted solution]

Hi @ugyentashi ,

Since 

103.7.116.143

is part of the directly connected subnet, R2 sends an ARP request for it and the FW replies with an ARP reply.

Regards,

[ugyentashi]

That makes sense, thanks. Just one query, what if the

103.7.116.143

was available(through a static route) through another interface on R2? ARP request will go through connected but server is available through another. will this create an issue?

[Harold Ritter]

Hi @ugyentashi ,

If 103.7.116.143 was reachable via a more specific static route, packet would be forwarded via the corresponding next hop, as more specific route would be preferred over the connected route.

Regards,