I've read numerous articles on Cisco's website, the support forums, whirlpool forums, dslreports & others trying to understand how PBR, route-maps & default routes work. Here's the deal:
ISR with NAT overload with the following iface's:
Dialer1 - WAN A on Gi0/0, dynamic
Dialer2 - WAN B on Gi0/0, dynamic
Fa0/0 - no ip address but 2 dialers mentioned above
Fa0/1 to Cable (DHCP, dynamic) - WAN C
Fa1/0 (routed port) with subifs:
fa1/0.10 & 0.20 should go to WAN A
fa1/0.30 should go to wan B
fa1/0.40 should go to wan C
I do NOT wish to implement failover or load balancing at this time. I am trying to wrap my head around PBR on this job but cannot make it work. I tried setting a default route to the cable provider, which works when I try to ping an external host from Fa0/1, but inevitably breaks my default route to the DSL dialers. I originally thought I could leave "ip route 0.0.0.0 0.0.0.0 Dialer1" and just worry about setting an ip policy route-map on the .30 & .40 subifs, and doing multiple ip nat overload statements with the appropriate route-maps.
Can anyone post a proof-of-concept config where I could get a better grasp at doing this? It can't be THAT complicated...
It's important to take note that I am running EIGRP on this machine because this router connects to a DMVPN network, in case that changes anything. I want the source interface of that tunnel to be WAN C. In fact, I might just end up putting Tun1 source interface to Fa0/1 and do no NAT at all, but even trying that won't work as there is no route configured in the ISR to Fa0/1's default GW.