ArubaOS and Controllers

Reply
Occasional Contributor II
Posts: 17
Registered: ‎09-14-2010

Controller discovery oddity

When AP's boot, do they *always* go through the complete sequence of controller discovery (dhcp option 43, multicast adp, broadcast adp, dns query)?

For example, if you provision an AP using the GUI (AP Installation), and enter "aruba-master" for "host controller" under "Master discovery", the ap is provisioned with the parameter "provision-ap server-name aruba-master".

Will this AP still go through all those steps of controller discovery or will it only try querying DNS for aruba-master?

Second, if the DHCP scope provides multiple name servers, and the first is unreachable, will the AP try querying a 2nd name server for aruba-master?

I ask because I had just set up a site with a local controller and things were working well. Suddenly the AP's went down and the only culprit I know about was that access to the name servers in their DHCP scope were cut off (a sys admin applied an erroneous firewall policy on the server's software firewall).

The AP's were terminated against the local controller, but provisioned as per the above. DHCP provided a domain suffix and two name servers, and querying for aruba-master.domain.suffix resolved the master controller.

While at the moment I did not have console access to the AP's, I could see from the switch ports they were patched to that the AP's were continually rebooting (the AP's were connected to non-Aruba/non-controller switch ports).

Would lack of DNS cause an AP to reboot? And even if DNS went down, some of the AP's were on the same VLAN as a controller interface--shouldn't an ADP broadcast have located the local controller?

Eventually we got one of the name servers back--the one listed 2nd in the DHCP scope. The AP's kept rebooting. It wasn't until access to the 1st listed name server was restored that the AP's finally booted and came up correctly.

That one's a puzzler for me and I wonder what happened. I suppose it's possible a network reconfig cut the AP's access to their LMS IP and backup LMS IP and that's why they rebooted. No one fessed up, though. All I heard about (and witnessed/experienced) was the lack of access to the name servers.
Guru Elite
Posts: 21,037
Registered: ‎03-29-2007

Re: Controller discovery oddity

If APs lose contact with the controller, they will rebootstrap and try to contact controller on the previous ip address without doing any resolution. If it STILL cannot contact the controller it will do a full reboot and then attempt DNS resolution, ADP, etc. Please see the knowledgebase article here: http://kb.arubanetworks.com/cgi-bin/arubanetworks.cfg/php/enduser/std_adp.php?p_faqid=61 that goes into detail about this.

Lack of DNS should NOT cause the AP to reboot. Loss of connectivity with the master controller would, however. In either event if an AP is on the same subnet as the master controller, it should be able to discover it, layer2 and reconnect to the controller's switchip.


Colin Joseph
Aruba Customer Engineering

Looking for an Answer? Search the Community Knowledge Base Here: Community Knowledge Base

Occasional Contributor II
Posts: 17
Registered: ‎09-14-2010

Re: Controller discovery oddity

thanks, Colin. How you and the kb article describe the process is how I thought it went...then I saw that strange behavior. Evidently there was a lot of extra work going on around me that I did not know about.
Search Airheads
Showing results for 
Search instead for 
Did you mean: