Wireless Access

We have found a problem where APs that have been down for a very long time (like months) seem to disappear from the down AP lists. In our case, it's because I had pre-provisioned a lot of APs prior to deployment, and so they had been sitting idle for a couple of months. They don't appear in the GUI list of down APs, nor do they show up with an SNMP poll. Has anyone else run into this? 

 

Also, this morning I plugged one of those "old" APs back in, and it came right up with its config and name, so clearly the controller remembers it. And when I take it back offline, it's now listed as down again. 

 

Maybe something with the "down" AP polling in the software is unable to read downtime or last up time beyond a certain range? 

 

(Currently have a TAC case open, but thought I would see if anyone else has already done this work.)

All an access point needs is to have the correct ap-group when it comes up and it will work.  If an access point address comes up and then goes down, it should show up in the Down list, unless it is cleared.  There are quite a few ways to clear down devices either by mac address or all the down access points on a controller, and that might have happened in your case.

 

http://community.arubanetworks.com/t5/Command-of-the-Day/COTD-clear-gap-db-access-point-housekeeping/td-p/1774

 

The controller does not proactively attempt to query downed access points.  It only reflects the status of access points that actively contact it.  Access points do not have an SNMP interface, so they cannot be queried via SNMP at their last known ip address...

Well, that's the problem. We have never cleared the config on these old APs, and they are using ap-groups that are still in existence and have not been changed since these APs were provisioned.  So they were up, went down, and then eventually just vanished from the lists. I've never done any housekeeping, though there was a software upgrade during that time. 

 

I understand that we don't contact the APs directly, we're trying to poll the controllers using SNMP. 

 

The issue for us is that we need to be able to know that the information we receive on our down APs is accurate. Without knowing why some of these APs are not being listed, then we cannot know with confidence that the information being provided either by SNMP or the GUI is accurate. 

 

From that link:  "By default, any AP that has booted to a controller creates an entry on that controller and the master controller's database, forever. If that AP is no longer in your system, it still shows up in the AP Down number on the master and local controller"

 

This is not what has happened. Those APs are no longer showing up in the AP Down numbers...

Support will tell you that the challenge is to ensure that those commands or the "Clear" link next to downed access points has never occurred.  Without seeing the audit trail for the entire time that the controller has been up, I am not sure what we can do, besides send syslog to an external server so that we can track the possible deletion of those access points.

 

Well what's frustrating about this is that the controller clearly "remembers" the AP. When it powers on, it receives the correct name and configuration information. My understanding is that these are all lightweight APs (AP-105) and retain no configuration of their own.

 

So I have APs that the controllers know about, that have not contacted the controller and are unreachable as far as the controller knows. Why then can the controllers not "know" that they are down? There is some kind of data mismatch going on here that results in inaccurate reporting. 

 

I don't like inaccurate reporting.

 

What AOS code are you running ?

Where are you looking to see if the AP is down ? Master or Local ?

What kind of controller setup you have ? Master-local , Standalone master ?

 

Airgroup AOS 6.1.3.6 (Upgraded from standard 6.1.3.4 at the behest of TAC)

 

Master with two locals. 

 

Polling all three of the controllers using SNMP to OID 1.3.6.1.4.1.14823.2.2.1.5.2.1.4.1.19

 

If you take a foreign or new access point and provision it offline with the correct access point group, it will receive the configuration from the controller for the ap-group and just work. There is no memory involved.

The APs were provisioned online, with the existing controller setups. The only difference between them and one that I might provision today is the amount of time they have been offline. Were I to take one and provision it today (as I have actually done this morning), it will show as Down when I unplug it. If I take one of these old ones out, plug it in, let it boot and come up, then unplug it, that AP will be listed as down. Its bretheren sitting in the box for months are not listed as down.

---

@cjoseph wrote:
If you take a foreign or new access point and provision it offline with the correct access point group, it will receive the configuration from the controller for the ap-group and just work. There is no memory involved.

---

So, the AP stores the ap-group configuration locally within itself, and then requests that config from the AP? That differs from how I understood it previously. Not sure how that affects the issue I have though.

---

@TESC-DanScherck wrote:

---

@cjoseph wrote:
If you take a foreign or new access point and provision it offline with the correct access point group, it will receive the configuration from the controller for the ap-group and just work. There is no memory involved.

---

So, the AP stores the ap-group configuration locally within itself, and then requests that config from the AP? That differs from how I understood it previously. Not sure how that affects the issue I have though.

---

That is correct.  The ap-group is stored in the access points' flash when provisioned.  It is presented to the controller when the access point finds it, and the controller send the configuration to the access point based on the ap-group that is presented.

 

Your issue, if I am understanding it correctly, is that access  points that are taken out of service show up in the "Down" list, but mysteriously disappear after time.  Please let support troubleshoot this and see if it is an issue with your setup.  Down access points should stay in the down list unless they are removed.

 

 

Ok, that's what I needed to know cjoeseph. (That this behavior is not in line with expected behavior.)

 

I will post whatever I find out from TAC.

 

Thanks!

What was the response of TAC ???