Excessive MAC Flapping causing slowness on Wifi Network using ZoneDirector 3000 and Cisco 2960S

  • 1
  • Question
  • Updated 2 years ago
Hi

We currently have a Wireless Network setup consisting of Ruckus ZoneDirector 3000 (using latest 9.9 firmware) and ZoneFlex Model 7982, 7762 and 7363.

The Core Switch is a Cisco 4507 R+E with a collapsed core design. Access Switches are 2960S connecting to the Core using Fibre Optic Cables configured as Trunkports. The AP's are connected to the 2960S Access Switches.

The Core Switch is doing the Routing between VLANS and the port configuration between Core - Access Switches - Access Points are identical which are listed below

switchport trunk encapsulation dot1q
switchport trunk native vlan X
switchport trunk allowed vlan X,Y,Z
switchport mode trunk

We are seeing excessive MAC FLAP notifications on the switches "show logging" command which we are suspecting is the cause why some of the AP's are going Up and Down. The Network is also sluggish when these MAC FLAPS are occurring. No other syslog messages can be seen on the switch, only MACFLAP and ofcorse if the interface becomes down or up.

A sample of the MAC FLAP log on the switches is shows below:

Mar  2 09:24:42.773: %SW_MATM-4-MACFLAP_NOTIF: Host 6420.0cbb.672b in vlan Y is flapping between port Gi1/0/3 and port Po3
Mar  2 09:24:43.156: %SW_MATM-4-MACFLAP_NOTIF: Host c462.ea85.c4f7 in vlan Y is flapping between port Gi1/0/3 and port Po3
Mar  2 09:24:43.193: %SW_MATM-4-MACFLAP_NOTIF: Host a088.b4e6.5f80 in vlan Y is flapping between port Gi1/0/3 and port Po3
Mar  2 09:24:45.143: %SW_MATM-4-MACFLAP_NOTIF: Host 6caa.b314.8590 in vlan Xis flapping between port Gi1/0/3 and port Po3
Mar  2 09:24:45.783: %SW_MATM-4-MACFLAP_NOTIF: Host 6caa.b315.4370 in vlan X is flapping between port Gi1/0/10 and port Po3
Mar  2 09:29:17.224: %SW_MATM-4-MACFLAP_NOTIF: Host 6caa.b314.a1e0 in vlan X is flapping between port Po3 and port Gi1/0/11
Mar  2 09:29:17.224: %SW_MATM-4-MACFLAP_NOTIF: Host 6073.5c80.8d3f in vlan Y is flapping between port Gi1/0/10 and port Gi1/0/11

On this particular Area having these excessive MAC FLAPS there are (2) access switches that are 24 Ports with 13 ports used to connect to ZoneFlex AP's.

Is there some configuration adjustments we can do on the ZoneDirector to help resolve this issue? Or any best practice configuration we can use on the Cisco Switches using Ruckus Wireless Access Points?

We also have a similar setup with our other clients and we are experiencing the same problem which we are still trying to resolve.

Please advise.

TZ
Photo of Tanski

Tanski

  • 5 Posts
  • 0 Reply Likes

Posted 3 years ago

  • 1
Photo of Munish Dhiman

Munish Dhiman, Employee

  • 100 Posts
  • 14 Reply Likes
Hi ,

 There is going to be a point during a roam when the MAC shows up briefly on two different switch ports. Clients will appear from different ports as they re-associate to other APs/cells in the ESSID. Are you sure this is causing aps reboot ?

Thanks 
Munish
Photo of Rahul Koul

Rahul Koul, Employee

  • 75 Posts
  • 12 Reply Likes
Hello TZ,

If you have "mesh" enabled on ZD and are noticing that some of the APs are changing their mesh states quite frequently then it may also cause the issue that you are seeing.

In order to avoid having the APs change their mesh roles you can go to ZD UI, Configure :: Access Points :: Mesh mode and set "Root" or "Mesh" mode. This will not let the APs change their mesh states.  Please apply this change only during a maintenance window as it causes APs to reboot.

I hope this helps.

Rahul
Photo of Tanski

Tanski

  • 5 Posts
  • 0 Reply Likes
Hi Rahul,

Yes we do have Mesh enabled right now.

We will do as what you have suggested and let you know if it solves the issue.

Thanks and Best Regards,

TZ
Photo of Tanski

Tanski

  • 5 Posts
  • 0 Reply Likes
Hi Munish,

Thanks for your reply.

For now we are not ruling out the MAC FLAPS as the possible cause of the AP reboot and sluggish network access as so far its only the logs we see on the switches.

Can you suggest other things we need to look at to isolate the issue that we are experiencing?

Just to add that with regards to spanning-tree, the Core Switch is defined as the Root Bridge with Spanning Tree BPDU Guard Default enabled on all switches. All Access Switches are connected to the Core Switch with single line fiber (no redundancy link) except for the one location which we have 2 links we bundled as Etherchannel.

The ZoneDirector was enabled as meshed as we are using mesh AP some locations without available cabling.

In another setup we have which is similar with the one I described above, we are seeing that whenever a syslog was generated for MAC FLAP on the switch, we noticed that on the POS (Point of Sale) Unit the response becomes sluggish, almost unusable at the time it occurs.

Thanks and Best Regards,

TZ
Photo of Bill Burns

Bill Burns, AlphaDog

  • 203 Posts
  • 38 Reply Likes
Tanski:

I had the same issue.
Things would work fine for a while but if/when there was a spanning-tree event (interrupting connectivity to the APs) the APs would attempt to restore connectivity by going into a mesh mode.
This caused loops, confusing the switches and resulting in the problem/behavior you described.

I disabled the mesh feature on every AP that was not intended to participate in mesh.
All APs that were intended to mesh but were connected to a switch were configured as "root"

Since making those changes I haven't had that problem.
There are still "mac flapping" messages that result from rapidly roaming clients, but there are a lot less of those messages, there are no strange topology loops, no more unexplained CDP entries and no performance problems.

FYI:
Be careful when upgrading firmware versions.
At some firmware version the R700 APs gained mesh capability and then I had to disable mesh mode on all the R700s to prevent this problem from happening again.