Wireless Access

 View Only
  • 1.  Issue with Controllers in active/active cluster terminating users traffic

    Posted Dec 25, 2024 06:28 PM

    We have two controllers initialized on md mode and clustered.

    There are 3 SSIDs broadcasted, and for all of them, the gateway on the users traffic is the vrrp ip of controllers in the associated vlan.

    The cluster is going well, both controllers are handling APs (around 90 per controller more or less).

    The issue is when both controllers are up, some users cannot access internet , even when they can ping the gateway. But if we turn off one controller, everything is going well. We know controllers are working in active/active way for controlling access point, but for the user traffic they are working in master/backup in each VRRP instance. Is there an issue in that architecture, with controllers acting as gateway of users traffic. We are thinking about asymetric traffic...

    How can we solve that issue, if we cannot put gateway of users traffic on firewall ?



  • 2.  RE: Issue with Controllers in active/active cluster terminating users traffic

    Posted Dec 27, 2024 01:46 AM

    Is there any reason why the Default Gateway of the users is pointing to the Warless Controllers VRRP? Are the controllers doing NAT?

    If not, i would recommend to put the Default Gateway on the Firewall and then you can easily check if its a Controller Problem or Network Miss-Configuration.



    ------------------------------
    Shpat | ACEP | ACMP | ACCP | ACDP |
    -Just an Aruba enthusiast and contributor by cases-
    ------------------------------



  • 3.  RE: Issue with Controllers in active/active cluster terminating users traffic

    Posted Dec 27, 2024 11:26 AM

    Hello Shpat,

    Thank you for your reply. 

    Actually, controllers are not doing any NAT. We are looking for a solution without using firewall as default gateway of users trafic, As because of hardware limitation, we cannot do it. Our Firewalls cannot handle traffic of around 2000 users a day.

    Regards




  • 4.  RE: Issue with Controllers in active/active cluster terminating users traffic

    Posted Dec 28, 2024 02:19 AM

    Hi  JKMendess,

    Can  you provide a topology or explain how the traffic flow is handled in your network? Question is, who does the NAT in your environment and who is providing the DHCP? Also, there might be a lot of questions coming up in order to understand your infrastructure and assist you with your concerns.



    ------------------------------
    Shpat | ACEP | ACMP | ACCP | ACDP |
    -Just an Aruba enthusiast and contributor by cases-
    ------------------------------



  • 5.  RE: Issue with Controllers in active/active cluster terminating users traffic

    Posted Jan 02, 2025 04:09 AM

    Do you have the user VLAN (L2) (and the uplink VLAN towards the network/internet) connected between the controllers? Are those VLANs and ports trusted on all controllers?

    Note that in your setup with routing on the gateway with VRRP, all active VRRPs should be on the same controller, and also all user traffic will run through that single controller. That's part of why it's deprecated to route user traffic on the controller, and using an external L3 (like a firewall) is the preferred option.



    ------------------------------
    Herman Robers
    ------------------------
    If you have urgent issues, always contact your HPE Aruba Networking partner, distributor, or Aruba TAC Support. Check https://www.arubanetworks.com/support-services/contact-support/ for how to contact HPE Aruba Networking TAC. Any opinions expressed here are solely my own and not necessarily that of Hewlett Packard Enterprise or HPE Aruba Networking.

    In case your problem is solved, please invest the time to post a follow-up with the information on how you solved it. Others can benefit from that.
    ------------------------------



  • 6.  RE: Issue with Controllers in active/active cluster terminating users traffic

    Posted Jan 02, 2025 09:32 AM

    In cluster operation, the controllers not only perform AP load balancing, but also client load balancing. This means that approximately half of the clients terminate at the VRRP master and the other half at the VRRP BACKUP controller. Presumably the clients from the VRRP backup also have routing problems. If you switch off one controller, the other will take over all client sessions and everything will work.

    You have to make sure that the client traffic between both controllers works in the corresponding VLANs. As Herman wrote, the interfaces and VLAN must be configured as trusted. Otherwise the controller will perform port authentication for incoming packets.

    Check on which controller the problem client terminates, use the command show user ip <client-ip> in the CLI of both controllers. This allows you to verify whether only the clients of the VRRP backup controller are really affected.

    Check what the controllers do with the client traffic, use the command show datapath session table <client-ip> and take a look at the flags.

    Are several routers involved in the data flow? If yes, which return route does the next-hop router use? It must also point to the VRRP IP address.

    In cluster operation, Aruba recommends handling all L3 services such as DHCP, DNS and routing on the upstream devices. The WLAN controllers should work like a switch and only establish an L2 connection between the clients and the gateway.



    ------------------------------
    Regards,

    Waldemar
    ACCX # 1377, ACEP, ACX - Network Security
    If you find my answer useful, consider giving kudos and/or mark as solution
    ------------------------------



  • 7.  RE: Issue with Controllers in active/active cluster terminating users traffic

    Posted Jan 14, 2025 10:00 AM
    Edited by jkmendess Jan 14, 2025 10:00 AM

    Hello Herman,

    Many thanks for your reply. And Sorry for the delay, i was out of office.

    I have already opened a TAC case. The support is actually looking a solution. They told me the configuration is ok. As you specified, all the interfaces are trusted, and the vlans are tagged between both controllers linked in L2 through a switch. On that switch we can see the users mac address on the differents vlans. I can also confirm you in the different VRRP instances created on both controllers for terminating users traffic, the same controller is MASTER in each one. But in the logs, it seems that a part of a trafic is handled by the second controller (BACKUP in the instance). It forwards traffic directly to the Firewall. But the reply of the Firewall is always sent to the CTRL which is MASTER. It seems we have asymetric routing with that configuration.

     For the moment one of both controller is disconnected, and all the user traffic is handled by one controller. Everything on that situation is working properly. the TAC is actually looking for a solution, as this behaviour is a non documented limitation for Aruba.