Wireless Access

last person joined: 20 hours ago 

Access network design for branch, remote, outdoor, and campus locations with HPE Aruba Networking access points and mobility controllers.
Back to discussions
Expand all | Collapse all

RAP failover process with ArubaOS 8 and clustering

This thread has been viewed 3 times

  • 1.  RAP failover process with ArubaOS 8 and clustering

    Posted Jun 15, 2020 04:59 PM

    Hi guys,

     

    How is the RAP failover process when running ArubaOS 8.5 and a cluster of MCs? I went through the "Aruba Remote Access Point (RAP) Troubleshooting" webinar and I saw the process in slides 19-21.

     

    https://es.slideshare.net/ArubaNetworks/emea-airheads-aruba-remote-access-point-rap-troubleshooting-80935505?from_action=save

     

    Slide 20 says "AAC fails and failure detected by S-AAC", and "AP tears down tunnel and S-AAC instructs AP to failover". In that situation I think there is seamless failover.

    However, what happens when the failure is not the AAC itself but the communication between the RAP and the AAC? Does the AP wait to miss a number of heartbeats from the AAC before failover to the S-AAC? If so, how many heartbeats? Then there is no seamless failover... Please your help...

     

    Regards,

    Julián



  • 2.  RE: RAP failover process with ArubaOS 8 and clustering
    Best Answer

    Posted Jun 15, 2020 06:52 PM

    The loss of communication leads to bootstrap. 

     

    However, this depends on the PAPI/GRE keepalive/retry mechanism settings.

     

    When communication is lost between the AAC and the AP (lets just say, a hello packet was sent to the AAC but the hello response got lost midway), This is usually detected through PAPI or GRE keepalives which can be configured as per the user's discretion.

     

    A GRE keepalive is sent out every 1 second, and the AP will bootstrap after 8 consecutive keepalive misses(assuming the bootstrap threshold is not modified).

     

     



  • 3.  RE: RAP failover process with ArubaOS 8 and clustering

    Posted Jun 15, 2020 07:02 PM

    Hi Ajay,

     

    Are these keepalive sent from the AP to the AAC and waits for the AAC response? Or are they sent from the AAC to the AP and waits for the AP response? Or both?

    And will the RAP reboostrap also after 8 keepalive misses as CAPs? I read somewhere RAPs will wait 30 seconds before failover, at least in ArubaOS 6.x, not sure if is the same in ArubaOS 8.x. Thanks for you interest.

     

    Regards,

    Julián



  • 4.  RE: RAP failover process with ArubaOS 8 and clustering

    Posted Jun 15, 2020 07:57 PM

    Are these keepalive sent from the AP to the AAC and waits for the AAC response? Or are they sent from the AAC to the AP and waits for the AP response? Or both? 

     

    The GRE keepalives are bi-directional.

     

    And will the RAP reboostrap also after 8 keepalive misses as CAPs? I read somewhere RAPs will wait 30 seconds -

     

    It is 8 seconds for CAPs and mesh portals, 30 for RAPs, 16 for mesh points. This can be configured ranging from 1 second to 60 seconds depending on the scenario. 

     

    However, since 8.5 GRE and ICMP keepalives are introduced.

     

    Some information on how that works can be found in the post below

     

    https://community.arubanetworks.com/t5/Controller-Based-WLANs/How-ICMP-keepalive-for-GRE-works/ta-p/657472

     

    Since a RAP is in the picture, this may be configured to actively check the WAN connectivity as well.