Wireless Access

I have a 6 controller network with a Master/Standby

Site 1, Local 1 and Local 2

Site 2, Local 3 and Local 4

At site 2, We have 2 VRRP instances on each,

Master on Local 1 for VRRP 1 and backup on Local 2.

Master on Local 2 for VRRP 2 and backup on Local 1.

We had a planned outage for Local 2 and moved the VRRP master to Local 1.

I confirmed all ap had moved over to the Local 1 and we where running for a few weeks in this state. 

When i rebooted Local 2 ( confirming no AP's where conencted with the show ap active), AP's running on Local 1 went down. 

Also we use ap-groups to direct AP to either local 1 and local 2 and about 50% of the AP in a group went down while the other stayed up.

So if a group has 10 aps, all connected to Local 1, on rebooting Local 2, 5 of those AP went down.

Why?

Why two VRRP instances?  The key in redundancy is predictability; do not mix and match where devices should be.  Dedicate a pair of controllers to only one set of access points and leave it at that.  Point the LMS-IP of the ap-group of your access points to the VRRP.  All of the access points should then be on the controller that has control of the VRRP.  Making it any more complicated than that will make things much more stressful during an already stressful time...

Hi, Thanks

Are you suggesting that all the AP should go to a single controller and the other controller is sat dormant. Why wouldn't you share the load between the 2 controllers and have the other as a back-up.

I thought it was a simple set-up. 

So each ap-group is a building on a site, so all aps in a group should terminate on a single controller.

Are we saying my set-up is an un-supported config or do i need to raise a support call?

I am suggesting that if you want to make things predictable, you should assign access points to a specific PAIR of controllers by pointing them at a single VRRP between them.  There is no advantage to load sharing unless your controllers are approaching the platform limits of access points, traffic or users.  

I am not saying what you are doing is an unsupported configuration, because I don't know the details, but if you want consistency, you have to design for it.  There are ways that you can design this that gives you the inconsistency that you are getting now.  There are also ways to design it so that it does not give you those results.  My post before favors consistency.

OK Thanks Colin,

I think i agree,I would like some predictablity.

Although I would still like to understand why rebooting a controller with no active AP's caused AP's connected to the VRRP pair to reboot.

What commands could I use to troubleshoot ?

Typically that should not happen.  I don't know enough about your setup to guess.  To find out why an AP rebooted, you would type "show log system 50" on the controller after the AP(s) comes back up and there might be a message saying why.  Alternatively the AP tech support would say why it rebooted last.