Wireless Access

 View Only
last person joined: 10 hours ago 

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

AP Failover to another cluster POC

This thread has been viewed 12 times
  • 1.  AP Failover to another cluster POC

    Posted Mar 22, 2019 11:32 AM

    Hello,

     

    Does anyone have any info on implementing fail-over for APs to another Cluster? Simply adding a LMS backup IP (Cluster 2 VIP) in the AP group didnt work :D I know a bootstrap would occur I don't even see the AP in MM after powering off the cluster.

     

    Additional info: I'm using DHCP options 43 and 60 for new APs.

     

    Thanks,

    Allan 



  • 2.  RE: AP Failover to another cluster POC

    EMPLOYEE
    Posted Mar 22, 2019 12:14 PM
    Are you powering off the cluster after the AP is connected to the cluster?


  • 3.  RE: AP Failover to another cluster POC

    Posted Mar 31, 2019 02:07 PM

    Yes I have two MCs in each Cluster. 

     

    I powered off each Cluster member. Watched the A-AAC change to the S-AAC. 

    Powered off the last MC of cluster. 

     

    AP never reported into BLMS (VIP) of Cluster 2.



  • 4.  RE: AP Failover to another cluster POC

    EMPLOYEE
    Posted Mar 31, 2019 02:17 PM

    When you say " you don't even see the AP in the MM" what command are you using the see the AP in the MM?



  • 5.  RE: AP Failover to another cluster POC

    Posted Sep 17, 2019 02:08 PM

    Is there a specific guide on how AP Failover to another cluster should be configured? i.e. where is the hierarchy the AP-Groups, VLANs, and WLANs should be created? I can find articles stating it can be done using LMS and B-LMS, but no suggestions on the actual configuration.



  • 6.  RE: AP Failover to another cluster POC

    Posted Sep 17, 2019 03:17 PM

    Do you want to make an AP failover to a specific controller in the cluster?

     

    For more information on how AP load balancing is configured, 

    refer the AOS UG 8.5.0.0 (Page: 373)



  • 7.  RE: AP Failover to another cluster POC

    Posted Sep 17, 2019 03:24 PM

    Say I have 2 different datacenters, and a cluster in each datacenter. If datacenter 1 cluster goes down I want the APs that typically terminate to that cluster to failover to datacenter 2 cluster.



  • 8.  RE: AP Failover to another cluster POC

    Posted Sep 17, 2019 04:27 PM

    So the scenario is like this.

    Datacenter 1 - cluster 1 - APs terminated on a controller in this cluster.

     

    Datacenter 2 - cluster 2 - APs should failover to a controller in this cluster if the controller that the APs are terminated to on cluster 1 fails.

     

    Try this solution

    Cluster 1 - This will have a MC which acts as a LMS to the APs (say 10.0.0.1 )

    Cluster 2-  Choose a controller which should act as a B-LMS to the AP when they fail over. (10.0.0.2)

     

    Assuming that the datacenters have L3 connectivity, configure a vrrp instance as follows.

     

    On Cluster 1 MC

     

    vrrp 1

    vlan <vlan used>

    ip address 10.0.0.1

    priority 110

    preempt

    authentication <choice of password>

    description Preferred-Master

    no shutdown

     

    On cluster 2 Preferred MC

     

    vrrp 1

    vlan <vlan used>

    ip address 10.0.0.2

    priority 110

    preempt

    authentication <choice of password>

    description Backup-Master

    no shutdown

     

    Now go to the AP system profile configured for the APs, can be viewed using the command "show ap system-profile"

     

    open the ap system-profile using the command 

    ap system-profile <name of the profile used>

     

    Configure

    lms-ip <cluster 1 MC>

    Bckup-lms-ip <cluster 2 MC>

     

    Save and deploy the changes.

     

    --Give Kudos: found something helpful, important, or cool? Click Kudos Star in a post.
    --Problem Solved? Click "Accepted Solution" in a post.



  • 9.  RE: AP Failover to another cluster POC

    EMPLOYEE
    Posted Sep 18, 2019 12:27 AM

    A_RAK is correct, that is the way of doing such a scenario. One thing to add. Instead of using a specific controller IP for LMS and Backup LMS I would use the Cluster VRRP for this entry. This makes sure, that the LMS or Backup LMS is available regardless of the status of a single controller. 

     

    The AP will first connect to the LMS IP. This should be the cluster VIP of cluster 1. it will then learn all Controllers in cluster 1 and will create tunnels to the AAC and S-AAC in Cluster 1. The Ap will only failover to cluster 2 if all controllers in cluster 1 are down or not reachable. The AP will not create a standby tunnel to a controller in Cluster 2 before all Controllers in cluster 1 are down. 

     

    Hope this gives more insights. 



  • 10.  RE: AP Failover to another cluster POC

    Posted Sep 26, 2019 12:35 PM

    Would you need to have all the VLANs configure on both clusters?

    e.g. Cluster 1 allowed VLANs configured on Cluster 2, and vice versa?



  • 11.  RE: AP Failover to another cluster POC

    EMPLOYEE
    Posted Sep 26, 2019 01:14 PM

    Yes.  OR, you could create the WLANs (Virtual AP) with a VLAN name and then define a name to vlan number mapping on each cluster.



  • 12.  RE: AP Failover to another cluster POC

    EMPLOYEE
    Posted Jan 07, 2022 08:19 PM
    Do you also have to configure the same WLANs, AP Groups, Roles, and Profiles on the Cluster 2 controllers? 

    If you enable pre-empt, does that take the head-ache out of moving the APs back to Cluster 1 after Cluster 1 comes back up?

    ------------------------------
    Jeremy Bradrick
    ------------------------------



  • 13.  RE: AP Failover to another cluster POC

    EMPLOYEE
    Posted Jan 08, 2022 01:47 PM
    Ideally, both clusters would be under the same folder, so they would have access to the same configuration.  If both clusters are not in the same folder/container, the APs will get the configuration for whatever ap-group they are in on the new cluster, which you would need to synchronize if you want the same behavior or you would have to manage two separate configurations.

    Honestly, I have seen few people fail over devices to a second cluster as a best practice.  Having a cluster is its own redundancy (APs would exhaust connecting to all controllers in a cluster before failing over).  Failing over to a second cluster provides more redundancy, but more complexity.

    Enabling pre-empt just makes things even more complicated, especially if you are not sure if only specific APs have failed over.  Long story short, your ultimate redundancy should just be a single cluster, unless you need site redundancy.  If you have a backup-lms pointing to a second cluster, that is where your APs will end up if they have failed to connect to all controllers in the first cluster.  Without pre-empt, that becomes your new APs cluster and they will "stick" to that cluster.  With pre-empt, things become more unpredictable as you will have to play the game of "where are all of my APs right now."

    ------------------------------
    Any opinions expressed here are solely my own and not necessarily that of Hewlett Packard Enterprise or Aruba Networks.
    ------------------------------