Thank you for the response Espen. This was actually the approach I took (flushing arp and reseting switches). It is a moot point now -- I came to the conclusion that I accidentally was spoofing the gateway interface using my Windows 7 MAC address.

Darwin award winner? I think so. I misinterpreted "Insert my local MAC address" in the Interface Edit screen. I thought it meant local to the interface I was editing. Not so! Lesson learned! My poor network was as almost as confused as I was.

However at this point I had not solved my original problem. I disabled my WAN interface just to see what would happen. This allowed me to ping my CentOS host. At that point it became clear to me that there was a routing issue -- taking down one interface causing another to start working seems like a "pecking order" issue to me. I had not checked the routing table before because the pfSense Wiki reads:
You do not need to add routes for networks which are directly connected to any interface of the firewall, and doing so may cause problems. You only need to define static routes for networks which cannot be reached via the default gateway.
I made the incorrect assumption that this statement implied that somehow no superfluous routes would be added, or if they existed they would automatically be removed. However for some reason it was configured to forward 10.144.1.8 to my WAN interface.

A quick route del -host 10.144.1.8 and my network is 100% functional.

However, still one problem remains. The route del command is not persistent when I reboot. How do I get rid of it? System>Routing>Routes indicates that no static routes are set up. Is there a routing configuration file somewhere?

Best Regards,
-Stefan

On 7/11/2014 6:35 PM, Espen Johansen wrote:

Please provide a network drawing.
I suspect you have a arp leak or a switch that needs to be restarted to clear its arp cache. Restart switche (s) without nothing connected and add the cetos and pfsense only and only after you have cleared both units arp cache (arp -d). Then take it from there.

HTH.

- LSF

11. juli 2014 21:57 skrev "Stefan Maerz" <[email protected] <mailto:[email protected]>> følgende:

    On 7/11/2014 2:03 PM, Stefan Maerz wrote:

        On 7/10/2014 7:52 PM, Stefan Maerz wrote:


            Hi everyone,

            I have a problem I have been unable to solve all day
            (literally *all* day).

            My pfSense box has two LAN interfaces and a WAN interface.
            A CentOS 7.0 server is giving me grief on one of the
            Subnets when configured as static or dynamic.

            When I put the problematic CentOS box on the other subnet
            (and change corresponding host network configurations), it
            works. The CentOS box also works when I put it on my
            trustworthy Linksys WRT router (again, changing host
            network settings along the way). To me this smelled of a
            firewall problem, but there is nothing logged and I have
            both LAN interfaces set up to pass everything. Secondly I
            looked at DHCP for possible DHCP addressing conflicts, but
            the DHCP server is disabled on this subnet. TCPdump
            reveals that literally nothing is making it to the gateway
            interface, however at the same time the activity light on
            the interface blinks corresponding to my pings (there is
            no other traffic).

            Further confusing me is that I am able to get a static IP
            from other devices when I plug them into the problematic
            subnet. Basically this single device does not work on this
            single subnet and that is the only problem. Other devices
            are fine on this subnet and this device is fine on other
            subnets. ...?

            It is also worth noting that all the link lights are
            lighting up and the cables and switch have been tested to
            be working correctly. Nothing that I can see looks out of
            place in pfSense's logs.

            Here are my host configuration files, all generated by
            CentOS's nmtui utility. I tried my own manual
            configurations with the same results (not
            working):http://pastebin.com/HFYYTG09(possible
            <http://pastebin.com/HFYYTG09%28possible> typos -- this is
            hand written, my apologies if that is the case)

            I am at a loss and have been at this all day. pfSense has
            so little to configure that I'm not really sure what I
            could have done wrong. I feel like it is something really
            simple that I missed. Anyone have recommendations on how
            to troubleshoot?

            Best Regards,
            -Stefan

        Hello again,

        I have been looking around. I found something interesting in
        my ARP table. For some reason my Windows 7 Admin workstation
        has two IP addresses assigned to it. Stranger yet is that one
        of the addresses a) is not on the correct subnet, and b) is
        the gateway address for the SRV subnet (the problematic subnet).

        Additionally there is nothing in the problematic CentOS box's
        ARP table.

        To me this would indicate that CentOS is trying to reach its
        gateway but pfSense is telling it that its gateway is on the
        wrong subnet (my Windows Admin Computer).

        I unplugged everything from the SRV (problematic) subnet and
        unplugged the ADMINISTRATOR computer then manually flushed
        pfSense's ARP table (arp -da). This removes the 10.145.1.38
        entry, but the incorrect 10.144.1.1 entry is persistent.

        How can I fix this? Is something misconfigured? Am I
        misunderstanding something? Is it a pfSense bug?

        See pictures:
        https://db.tt/4Q6skpiF -- pfSense ARP table (some data sanitized)
        https://db.tt/coWT9GoB -- Settings on ADMINISTRATOR PC

        I am greatly appreciative of any help.

        Best Regards,
        -Stefan

    I believe I have indeed confirmed something is indeed wrong with ARP.

    Here is a TCP dump session from the pfSense Box*:
    http://pastebin.com/zyVpPkcD

    As you can see it is looping through ARP. Line 6 is pointing my
    CentOS box at a MAC address on a separate subnet.

    So I suppose my question is now, how do you troubleshoot ARP?
    Isn't it supposed to just work? What am I doing wrong?

    I also have a remote site that could benefit from this knowledge*:
    https://db.tt/5xKvBZBb

    Best Regards,
    -Stefan

    *Data is sanitized
    _______________________________________________
    List mailing list
    [email protected] <mailto:[email protected]>
    https://lists.pfsense.org/mailman/listinfo/list



_______________________________________________
List mailing list
[email protected]
https://lists.pfsense.org/mailman/listinfo/list

_______________________________________________
List mailing list
[email protected]
https://lists.pfsense.org/mailman/listinfo/list

Reply via email to