I recently inherited a campus-wide mesh network of 6 RAPs and 14 MAPs. We have 4 remote switches bridged through MAPs. As-is, the network is working, but there are some things that we would like to clean up and future-proof on the LAN.
In their current state, most of the remote switches are using only 1 VLAN. We would like to separate this traffic as we do in the rest of the network. I've been having issues getting trunks set up over the air. As many have said, Cisco's documentation is very confusing. I've gained some insight from Scott's powerpoint, but I'm still having trouble making it work. Here is a sample portion of what we have and what we are trying to accomplish. (All management interfaces are in VLAN 95)
Current state:
[CORE] ->[SWITCH-NATIVE95] -> [RAP(Interface Normal)]~~~~~~~~~~~~[MAP-(Interface Normal)] -> [REM-SWITCH-NATIVE95]
In this state, anything in VLAN 95 connects to the rest of the network, but anything outside of VLAN 95 does not. If I flip the MAP's interface to trunk, with a Native VLAN of 95, I can ping everything in other VLANs, but lose connectivity to the switch.
This is what I tried today, with 95 and my other VLANs as allowed VLANs (on the switch and the MAP):
[CORE] -> [SWITCH-NATIVE999] -> [RAP(Interface Normal)]~~~~~~~~~~~~[MAP-(Trunk-Native999)] -> [REM-SWITCH-NATIVE999]
This also resulted in no connectivity to the switch. I also noticed that the MAPs that normally connect to this RAP all chose to connect elsewhere.
I have read that RAPs' interfaces should always be set to normal and you will get a warning when trying to set it to trunk. I attempted to set it to trunk and got no warning, so I tried this:
[CORE] -> [SWITCH-NATIVE999] -> [RAP(Trunk-Native999)]~~~~~~~~~~~~[MAP-(Trunk-Native999)] -> [REM-SWITCH-NATIVE999]
I got the same result as with the interface set to normal. Also could not get the MAP to reliably connect to the same RAP to do a controlled test.
Questions:
What am I doing wrong? How do the interfaces have to be set to trunk over the air?
I was consoled into [REM-SWITCH] and didn't get any VLAN misconfig messages, but I did notice the RAP going up and down intermittently via NCS. Could SpanningTree be an issue?
Does the AP's management interface have to be in the native VLAN?
Is there any way to force a MAP to connect to a particular RAP?
(Cisco AIR-CAP1552I Access points, WLC-5508 ver 7.2.111.3)
