Reference this drawing:
http://www.butchevans.com/readarticle.php?article_id=16
I have this set up on my WISP like this:
T1-router – RB532 as Router A – two pairs of WDS links – PentiumIII-PC as Router B – switch – customer router – switch – customer
All assigned IPs are public except on the WDS links (they are given RFC 1918 IPs for management purposes). traceroutes show public IPs all the way … like this:
from outside to T1 to customer:
T1’s IP, Router A (eth3), Router B (eth2), customer router IP
from WISP to T1 to outside:
customer router, Router B (eth3), Router A (eth1), T1-router-eth, T1, the world
Physically, it goes like this due to OSPF programming as in Butch’s article (er, well, it’s supposed to):
T1, eth3-Router A, eth1-Router A, eth1-Router B, eth3-Router B, customer-Router, eth3-Router B, eth2-Router B, eth2-Router A, eth3-Router A, T1
The IP addresses on my “link 1” are different from the addresses on “link 2”. They have to be for this OSPF/'full-duplex" thing to work. Notice that all my eth1’s are with “link 1” and all my eth2’s are with “link 2”.
All packets that enter my WISP network (the customer) traverse “link 1” and all packets that leave my WISP network traverse “link 2”. It seems that everything works fine except:
Two issues:
When a customer goes to “http://whatismyip.org” they see only the eth3-Router A address, not their real one at their actual router.
I’m thinking I do not have my routing 100% correct on Router A and/or Router B. Can anyone help out here?
Also, supposedly, IPSec will not work from my WISP network to an outside IP address.
Does IPSec need the same physical path (in my context above) or same IP path to work properly? Customer says that VPN “comes up but won’t pass any data”.
I don’t ask questions very often. I hope someone can lend a hand here…
Thanks,
Michael