12-19-2012 02:50 PM
Over the last year I have noticed several instances (usually after power outages) where a number of APs would show down on the master controller (GUI and CLI) but they would actually be up and working as intended. We've upgraded to the 18.104.22.168 code but that hasn't seemed to make a difference. As a quick fix, I usually just reboot the APs in question and they show back up on the master after they boot.
The APs that do this are mixed model and they're in no particular subnet, vlan, or groups. It seems to be fairly random. Our current model consists of two Local controllers that handle all of the AP terminations and one Master above them.
To clarify, this happens after the APs lose power during an outage and come back up, not the controllers.
Is there any way to force the Master controller to see these APs without rebooting them? I can ping these APs from the Master and I tried "clear gap-db ap-name XXXX" but that simply removed the down entries. This hasn't been causing any issues for the end-users but it can be a headache from the administrative side of things.
12-19-2012 03:39 PM
22.214.171.124 release notes, under known issues -
"A subset of APs connected to a local controller are up on the local but are marked as down on the master controller. This occurs when campus APs (CAPs) on master or local controllers, not using control plane security, discover the master using DNS, DHCP, or ADP and connect to the master before moving to the local controller configured in the AP system profile. An entry is created on the master and local controllers simultaneously. When the AP is aged out on the master, it is marked as down.
Workaround: Execute the command clear gap-db lms <lms-ip> on the master using the IP of the local controller. This should make the LMS send an update of all the APs marked UP and the master should show the correct status. Additionally, this issue can be resolved by rebooting the APs."
Aruba Customer Engineering
Looking for an Answer? Search the Community Knowledge Base Here: Community Knowledge Base
12-19-2012 04:42 PM
Thanks a bunch. Normally, I'd only see a handful go down like this but we had about 100 showing down after the last outage. This is very helpful to know. I found a previous thread about this issue but the command was seemingly redacted from the last post.