Statically set APs constantly dropping onto DHCP

[m.s.rees1]

Apologies if that has been brought up previous but I couldn't seem to find anything on it.

We're currently running three WLC with over 1000 APs joined and being monitored etc. All the APs are set statically and work fine, randomly however but quite often an AP will be forced over to DHCP and change its IP address. Highly annoying when they are being monitored via the static addresses and our monitoring software constantly flags 'down' access points when they are in fact up. Once the AP is rebooted it will jump back on the static address with no problems. I'm trying to work out why this is happening in the first place. Any help would be much appreciated.

Update: I have noticed that the high availability information set on the controllers for an AP are not 'applying' themselves on the AP itself. When I telnet into an AP it can often have no controller information or the wrong controller (if it was moved through planned maintenance, it still holds the previous controller though the high availability information says otherwise), could this be a key factor? 

WLC: 2 x Cisco 5505 and Cisco 8500 (Happens on all three controllers, all firmware recently updated)

WAPs: 1042n, 1602i 

Please see screen shot of log of an AP that has done it recently.

Thanks

[Rasika Nayanajith · Accepted solution]

I have noticed that the high availability information set on the controllers for an AP are not 'applying' themselves on the AP itself. When I telnet into an AP it can often have no controller information or the wrong controller

This is normal & global backup controller configuration is not propagate to AP it self. You have to configure AP High availability with primary, secondary, tertiary WLC with name & IP if you want predictive failover for a given AP. Refer below for more detail

http://mrncciew.com/2013/04/07/ap-failover/

if it was moved through planned maintenance, it still holds the previous controller though the high availability information says otherwise), could this be a key factor? 

If you haven't configure HA field on AP, still it remember all available WLCs within your environment & try to register one of them if AP went through a power outage. But no preference where it will go unless HA field configured on that AP.

Regarding static AP forcing to DHCP, it is also part of WLC discovery process. Due to some reason if AP registered WLC was unavailable, it will try to discover another WLC. After different method of discovery, AP will power cycle & start the discovery process again. Discovery via DHCP is one option. Here is some basic information about AP registration

http://mrncciew.com/2013/03/17/ap-registration/ 

I would try to reserve IP address via DHCP for these AP. In that case AP should get the same IP via DHCP & your monitoring can be done as it is.

HTH

Rasika

**** Pls rate all useful responses ****  

[m.s.rees1]

Helpful information to know, thanks Rasika. So if the information on the AP itself is set to the wrong controller (or has no controller set) with no secondary controllers set, could this be one reason that it is jumping across to DHCP occasionally, to find a controller?

[Massimo Baschieri]

I've done some quick testing and IT SEEMS that if you disable dhcp server ap's stay quiet on their static ip, anyone can do the same and confirm or not that behaviour?

If it's confirmed to work, that workaround is feasible however only in case you have all ap's statically configured.

You can eventually manage maintenance windows with active dhcp server in case, for instance, of new ap's setup.

[m.s.rees1]

We'd rather not disable DHCP to be honest but good to know. Thanks for helping with this Massimo. I'm starting to see a potential pattern with the AP's that are doing it. It seems that the only ones that are doing it are ones that haven't had the primary controller configured on the AP itself. I'm continuing to monitor this to see if this is the link. I haven't come across an AP that has done it that has the correct controller set up in it, so we'll see.

I'm not sure if the AP's are checking the controller occasionally and failing to get back to it because of the wrong or lack of controller configuration, resulting in it falling on to DHCP.

Thanks again.

[Massimo Baschieri]

As far as I can recall I've already tried with brand new ap's with no primary controller configured (dhcp provided wlc address), primary controller configured and primary controller configured then removed, but in all that cases ap's tried dhcp when wlc down.

Anyway, maybe I'm wrong, so please keep informed about your tests. 

[Scott Fella]

Some of my clients also use static address assignments and what happens is that if the AP doesn't find a controller in a given time, it will fall back to using dhcp. So a perfect test would be to move an access point to a different vlan and look at the console of the AP.  This feature was implemented as one of those, "just in case".  What you probably want to do is mac reservation on your dhcp server.  This was your AP's if for some reason try to dhcp, will get there same address. Since your monitoring via ip, I would use dhcp reservation and disable the static address. This should provide you with a more stable environment for monitoring.  Many of my clients do this so they control what ip and ap has.

-Scott

[m.s.rees1]

I think this would be a good solution however the implementation for us wouldn't be practical. We have around 1500 access points scattered across more than 200 sites, all with different IP ranges, scopes and vlans. This would take a while, unless we did it gradually. We only get around 5 APs dropping to DHCP a week, so I doubt this would be approved to be honest. Thank you for your input though, it's appreciated.

Just to clarify, the problem with us isn't the initial 'joining to the controller' process, the AP's will already have associated with the correct controller on the static and will work fine, but then randomly without no apparent reason (no power loss or anything) it will drop to DHCP and the monitoring system flags that it has gone down. If you go to the controller that the AP is on, it will be there and working fine, just with the wrong IP. If it is restarted or reloaded, it will come back, associate fine, and have the correct IP again (which is the static)

Update: I still haven't found an AP drop to DHCP when the correct controller is set on the AP itself manually. I'm not sure if this is the fix or if it just hasn't happened on one of these APs yet. I shall keep you updated if I find out any more information.

[Massimo Baschieri]

My experience is slightly different, a customer of mine lives in an area that sometimes suffer from long power outages, in such cases all the IT infrastructure goes down, since UPS's doesn't last so long.

When power comes back ap's will boot a lot faster than virtual wlc, that time gap is enough for ap's to start looking for dhcp and get dynamic addresses.

I believe I've already tried to manually configure the primary controller on ap's but nothing changhed.