Rebooting Meru 832AP's bring down entire network (wired and wireless)

[JHatz]

I have Meru 832e AP's in my High School - all connected at Layer 3. I use a Meru 3200 controller and its connected to my core switch which is Enterasys. The switch was put in about a year and a half ago - the Meru AP's were put in this summer ( I had 320i's previously - with no issues - but they were L2 at the time).

 

 

I am able to take down our entire network by simply rebooting three or more of the AP's - meaning wired and wireless connections no longer get to Internet or internal addresses - running pings gets maybe one response out of 20 or more if lucky.

 

To get my network back - in most cases I can simply unplug the Meru Controller's ethernet cable from the core switch - wait 30 seconds - plug back in - and after maybe 20 seconds the whole network is operational again. In some rare cases I've had to reboot the Meru controller. I can replicate this at will by simply unplugging or rebooting three AP's. If I do only one - it will take the whole network down for the time it takes four pings to be dropped - then things come back for five or six pings - go down again for four pings then things go back to normal.

 

I've had one instance when our network went down like this - except I didn't reboot any AP's manually - and we only lost Internet traffic - not internal connections as I could still talk to servers in our buildings. I tried unplugging the ethernet cord to the Meru controller and plugging back in - then I lost the internal network as well - I had to reboot the controller to bring things back to normal.

 

We've done some wiresharking on the controllers port - and don't see anything I'd call flooding - looks just like normal amount of traffic. Nothing is standing out in the pcaps. It almost seems like it could be ARP cache poisoning - but I don't know how to prove that.

 

Anyone else ever come across anything similar?

[njc235]

AS your system is still in support, I would suggest a call to Meru

[JHatz]

I've got tickets opened with both Meru and Extreme - neither have come up with anything yet thus why I decided to try here - was hoping someone else has gone through this sort of thing and has the answer.

[Ashm]

Could it be an issue with Spanning Tree? Is there a command on the Extreme core switches where you can check STP History to see if there has been any topology changes when the network goes down? And just check the logbuffer in general on the core switches to see if they point to anything obvious.