IPSEC One Way Traffic-ish (seems like a bug)



  • Hi pf gurus.

    I really hope you can help put me out of my misery.

    I am using 2.0-RC3.

    I have setup a IPSEC VPN between our office and our remote data centre.  The tunnels come up fine.  I have added (while testing) an IPSEC rule to accept everything.

    If I ping a host at the data the traffic leaves the lan via the vpn and all is well.
    If I ping a host in the office from the data centre then I am told they can see traffic leave via the vpn.  I can see it if I do a packet capture but it doesn't get thorough.

    If I then jump onto the host I can't ping and do a reverse ping (office to dc), that works AND the ping (dc to office) magically starts working.

    Any thoughts where I should look?



  • Firewall issue on office host?
    Do you have pfSense on both ends and are they the default gateway?
    If not "Bypass Firewall Rules for Traffic on Same Interface" (do a forum search ) and static routes might be needed.



  • pfSense in the office which is also the gateway (multi-wan) whilst the DC is running a Fortigate 200A.

    "Bypass firewall rules for traffic on the same interface" is not currently ticked.  Changing it made no difference.

    If I do a packet capture on the pfSense IPSEC if, I see nothing whilst the ping is not working.  As soon as the ping starts working (forcing it as above) then packets appear in the capture.

    I see nothing obvious in the firewall log to show anything is being blocked.



  • Google "asymmetric routing" as it must be the problem.
    To eliminate you could setup pfSense at home and make a site to site connection to the office.



  • While not having had the chance to setup a tunnel from home to test as suggested, I have built another install on a spare box at work.

    Vanilla install with minimum config (LAN and WAN etc)
    Single WAN, no load balancing, nothing.
    Setup the IPSEC Phase1 and Phase2s.
    Add rule to IPSEC if.
    Bring tunnel up.
    Test

    Result…  Exactly as described earlier.

    I will keep trying and post back any other findings in the hope it helps someone else.



  • I have no problem on a snapshot from jun 14



  • I can confirm that the problem WAS NOT my config or PF but in fact it was the data centre config and not managed by me.


Locked