I started off by contacting aws support, and they gave up and told me to ask here
The idea of the AWS transit network is that it uses BGP to negotiate the interconnectivity of subnets between worldwide regions, as there is no direct way of doing so within aws
We are trying to route traffic from 172.30.0.0/16 via the transit network, and onwards to others.
TCPDump shows packets leaving this network, arriving on the destination network, and the reply being sent back. However this reply never reaches there
Some digging around on the cisco CSR showed something rather strange when looking at the bgp info
Route Distinguisher: nnnnn:3 (default for vrf vpn-12345678)
*> x.x.x.x/16 y.y.y.y 0 9059 i
*> x.x.x.x/16 y.y.y.y 0 32768 ?
*> x.x.x.x/22 y.y.y.y 200 0 9059 i
*> x.x.x.x/24 y.y.y.y 200 0 9059 i
*> x.x.x.x/24 y.y.y.y 100 0 9059 i
*> x.x.x.x/24 y.y.y.y 200 0 9059 i
* 172.30.0.0 169.254.45.17 200 0 7224 i
*> y.y.y.y 100 0 7224 i
*> x.x.x.x/22 y.y.y.y 100 0 9059 i
* x.x.x.x y.y.y.y 100 0 7224 i
* y.y.y.y 100 0 9059 i
* y.y.y.y 100 0 9059 i
* y.y.y.y 200 0 7224 i
*> y.y.y.y 100 0 7224 i
I've blanked out a lot of the other details, but left the subnet masks in
My support contact at aws assured me that the vpn tunnel was sending its local network correctly, so it must be something on the cisco side of it. He recommended setting up a route map to filter out 172.30.0.0, but this does not seem to then allow 172.30.0.0/16
Has anyone got experience with manually manipulating BGP lists, or suggestions on a work around?
