Wireless Access

Current Scenario

We are currently running AOS 8.3.0.1 with 2 x Aruba 7240 controllers and 1572 x Access points + Clearpass + Airwave.

The current configuration of mobility controllers are in HA mode.

New Scenario

Ideally I would like for us to move into a cluster scenario using all 4 x controllers (7240’s)

We have two x 7240 controllers I am looking to configure and install this week and place in their own cluster. (Cluster B)

The idea is then to remove HA from the existing 2 x controllers and add them to their own cluster (Cluster A) so we can use master/standby VIP for AP Load balancing (LMS and Backup LMS)

Question

My question is as follow: If I add the new 2 x controllers to the mobility master and create the first cluster, will AP’s automatically start terminating on these two new controllers? (If so, is there a manual way to prohibit the AP’s from terminating on the two new controllers until the full migration has been completed i.e. we have removed HA from the 2 x existing controllers and re-configured them for clustering.

Also if I remove HA from existing controllers and apply clustering, will we experience any downtime? (Or is this the part where the 2 x new controllers in the 2nd cluster can come in handy?) Or can we just add clustering to existing config and proceed again with removing HA and applying clustering.

Same question remains which is very important: Will we experience any downtime and if so for how long more or less so we can prepare our change request/risk accordingly.

How are the APs currently discovering the controllers?  I assume ADP.

Install the new controllers with adp disabled so that aps don't appear on there and create the cluster. Create a vrrp between the two and point your aruba-master dns to that address.  When you reboot the aps they will then come up on the new cluster.

Hi MIchael,

That certainly helps with my question regarding the two new controllers and how to ensure AP's don't terminate on them - thank you.

My other concern is the current controllers that are in HA mode. I need to create a cluster for them and I guess it is quite straight forward i.e. remove HA and create the cluster but what I really need to know is:

1) Will the controllers require a reboot

2) I guess if the answer to 1 is yes, then AP's also need to reboot?

So with 1500 plus AP's across campus, how much downtime should I plan for or can I move AP's to new controllers and cluster and then reverse current controllers i.e. disable ADP on them whilst changing from HA to cluster and then enable ADP again.

Hi,

Removing the HA stuff won't in itself cause the controllers or APs to reboot but you may end up with some unpredictable results.

Best thing is to test the new cluster by moving a bunch of APs across to it.  You can point the aruba-master to the new vrrp or use an ap provisioning profile to point them to it (this will cause the aps to reboot). 

Once you are happy that everything is working as expected, go ahead and move all the aps across to the cluster.

Then remove the ha config on those two controllers.  You could then join them into the new cluster so that you have a 4 node cluster.