Load Balance not Working (2.0-RC3 (i386) built on Fri Jul 1 00:16:18 EDT 2011) ?

  • Edit: After numerous and millions of times reading this http://doc.pfsense.org/index.php/Multi-WAN_2.0, I think I got it right.
    I will answer my own questions in red color and you can agree or not to them.

    I know there are already a lot of posts regarding loadbalancing but please bear with me regarding my concerns.
    I cannot pinpoint what's wrong my setup. My concerns are those in bold text. Thanks for those who will give their expertise/advice.

    2.0-RC3 (i386)
    built on Fri Jul 1 00:16:18 EDT 2011

    I tried to loadbalance 2 WANs(WAN2=Static & Default, WAN1=Dynamic) with 1 LAN.

    But as you can see in the picture, there is a sudden drop on the RRD of WAN2. Sometimes, no traffic going to WAN2.
    Is this normal? Yes. If the LAN clients of WAN2 have no more traffic activities or if WAN2 has no LAN clients at all. If the LAN clients of WAN2 have no internet activities, the traffic graph should not be flat zero for IN. At least it should be at 1kbps more or less just like in the picture, the IN ranges from 2-3KBPS. Because if it is zero, then the interface is down.

    Squid is installed but not turn on.
    vnstat2 is installed and running.

    FW rules, only the basic.
    Use sticky connections and Allow default gateway switching are both enable.

    Is loadbalancing in the System>Routing>Groups is enough?
    **Or do I need to put my WANs' gateways in the Services>Load Balancer>Pool ?no need
    I don't know what are the correct ports in the Services>Load Balancer>Pool (please see attached image).remains a question

    Also in Systems>Routing>Edit Gateways>Advance section, there is weight.
    **If for example my WAN2 is my Default Gateway, will this be given a 1 weight or can I assign the weight of 1 to my WAN1 which is not my default gateway?**weight is independent of the default gateway. For example my WAN2 is my default gw and I give it a weight of 2 and my WAN1 has a weight of 1, then in the round robin algorithm, First traffic will go to my default gw which is WAN2. Second traffic will go to same gw WAN2 because it has the weight of 2. Meaning it will serve the traffic twice. The third traffic request will go to WAN1. Fourth will go to WAN2 again which is the beginning of another round robin cycle. Fifth will go to WAN2 still. Sixth will go to WAN1. And the cycle continues.

    The only concern remaining is this: kernel: arpresolve: can't allocate llinfo for {WAN1 GW}


  • for basic loadbalancing you just need to define gateway groups !! info about how to do this is found here: http://forum.pfsense.org/index.php/topic,10407.0.html

    remove all your settings in the load balancer pool menu's .

    in your firewall rules you should select the gateway_group you created as gateway, this will result in loadbalancing if you set all gateways to the same Tier

    loadbalancing with transparant squid is a bit tricky … for more info see: http://forum.pfsense.org/index.php/topic,37083.0.html

  • hi sir heper:

    yup, as what i put in red font, there's no need for the load balancer pool for the basic loadbalance. thanks for confirming it.

    for my concern in loadbalancing with transparent squid, I posted a reply to your link: http://forum.pfsense.org/index.php/topic,37083.0.html

  • I'm experiencing a similar load balancing issue, however I have been using the correct new method in 2.0 of using gateway groups.

    WAN-1 being testing from 2:00 AM until 8:00 AM:

    WAN-2 being testing from 2:00 AM until 8:00 AM on the same day:

    The connection just stops for an extended period of time, but only on WAN2. Whats interesting is that if you compare WAN1 with WAN2 it looks like something similar occurs on WAN1 at the exact same time that the issues on WAN2 begin and end.

    At first I thought it may just be an issue with the upstream provider for WAN2 however this has occurred over numerous nights of testing. I also decided to run an OpenVPN connection over WAN2 one night with extremely verbose debugging at both ends and it indicates that the connection is still active just very little data is being routed when the slow down occurs.

    To try and isolate the issue I then decided to test a configuration where WAN2 is used exclusively and no load balancing occurs.

    As you can see it is what is expected, a constant saturation of WAN2 over the entire period. I have performed further testing over other nights since and it seems the deciding factor is the load balancing via a gateway group. What is also interesting that it always occurs after approximately 1-2 hours of sustained load balancing and resumes actively load balancing over both gateways after around the similar amount of time has passed (i.e. each night during testing the same sort of RRD graphs were produced).

    I have changed the monitor IP on both WAN1 and WAN2 to both the actual gateway IP as well as an external DNS servers to see if the WAN2 gateway was being dropped from the group (this wasn't the case however).

    The issue I'm having looks very similar to the issue experienced by the OP (sudden drop in transfers over WAN2). But what is different in my case is that I'm using gateway groups as opposed to relayd balancing or anything such as that.

  • i have the same problem

Log in to reply