Hello list, I am experiencing a very strange behavior with policy routing using PF, i am using OpenBSD 4.8
The networking scenario is as follows 3 BGP Upstreams, 2 with multipath and 1 directly connected Lets say UPSTREAM1 = directly connected UPSTREAM2 = multipath UPSTREAM3 = multipath i am announcing a 192.168.10.0/21 to all upstreams and 192.168.10.0/24 only to UPSTREAM1, above this, i have set up a pf rule that matches ip source 192.168.10.0/24 and sets the next hop to UPSTREAM1, the rule is as follows pass in on vlan20 from 192.168.10.0/24 route-to (em5 UPSTREAM1_IP) no state now the problem, the networks that i learn from bgp and prefered from UPSTREAM2 cant reach their destination, but the routes that are prefered from UPSTREAM3 reaching their destination inside me, if i change the localpref of a network that now i prefer from UPSTREAM2 and make it prefer UPSTREAM3 it works, can any developer help me and tell me if there is any connection between pf's policy routing mechanism and the OpenBGPD or the system's global routing table at all? any suggestion on what to look? my last resort will be to reboot the machine :) I hope that i have explained the problem good, if you need extra informations please ask me :) thanks in advance (This router has 576 days uptime :) thank you OpenBSD for your good software)
