Mobility Express - Weird occurrence booting new APs in an existing ME subnet

[gp1200x]

We have probably 15 subnets each with a unique ME setup running on them with groups of APs in each ME. Each acts as a unique ME. ( I know only one ME is allowed per subnet.)

I preconfigured a bunch of APs (3802Is) on one subnet so they would have the lastest code and they were set using DHCP for the WAP IP address.  Usually if I just boot one in a subnet that already has an active ME they join in that ME instance and I then hardcode an IP on them (don't like using dhcp on running APs). I normally DO NOT even have to hit the reset button for 20 seconds for this to happen. All MEs were at 8.5.140 for the last year.

I upgraded sites to 8.5.151.

I placed a few preconfigured APs from another subnet into a new subnet that was already running ME with a primary controller and AP3802is. I did NO do a reset on them first.  To my surprise they did NOT join the local subnet but instead joined in the ME that was in a different subnet on the other side of the wan!  They came up using dhcp (no surprise there) but managed to find the ME instance running in another subnet (the one they were originally configured in) and joined into that ME group. TO MY surprise the Open Internet SSID WORKED which was a trunked vlan BUT NOT the correct vlan id for that site it was sitting in. The vlans IDs were different from the site it joined and the site it was residing in. Totally preplexed me but I was out of time to really look into how it worked and why it joined an ME running in another subnet. I realize it got the info for the ME controller address from its previous config but I never had this happen before.  A 22 second reset press on the access points reset them and they joined the ME in the local subnet after they rebooted.   

So my question is...was this normal because I never had it happen before (new code function??)  and can you actually have a setup with ME controller in one subnet supporting APs in other IP subnets?  Maybe it would work okay in a local site with multiple vlans .......but APs crossing over a wan link to communicate with the ME primary controlller???  you would most likely have a mismatch with different vlan ids on the trunked interface.  

Thanks

[pieterh]

ME controller discovery works the same as dedicated controller discovery

multiple methods are tried sequentially and after discovery is done a controller is selected.

Lightweight AP (LAP) Registration to a Wireless LAN

if a) and b) may not be configured on the new created subnet?

"preconfigured controller" comes before network broadcast on local subnet.

[Mikey Boy]

I think you have potentially deployed these in a scenario Cisco did not anticipate in that they have come from a site with an ME Controller and you have powered them up on a site with a different one that has routed access to the old ME controller.

Traditional controller based CAPWAP APs will always try and connect to their configured or previously joined controllers first.

The ME AP is basically running an AireOS operating system so it functions pretty much the same with some limitations. I suspect the AP you have powered up and just went straight to its last known controller rather than looking for a local one.

 As you said, performing a reset on the AP before powering it up led to it operating as you would originally have expected.