Strange behavior of two paired C25

  • 0
  • 1
  • Problem
  • Updated 2 years ago
  • Solved
Hello, everybody!


I have a very strange problem with Extreme Wireless deployment.

There are two C25 paired controllers (9.21) and 65APs (3705 and 3710), spread equally between two buildings: A and B. (Please, take a look at the attached scheme).

There is one open SSID everywhere on all APs and no any authentication or portals. All the clients get their IPs through DHCP server, configured on X670 core switch at Build A. Address space is /22 network.

When I use wifi inside a certain building (A or B) its ok - I can move between the floors, watch youtube, browse websites and there are no any interruptions. 

Strange things happen when I walk from one building to another. For example, I start to watch a long (~ 1hr) youtube clip in Building B on my laptop at 4th floor. I see that browser starts to cache the content of the clip. Then, I get to floor 2 of Building B and go to Building A through a long glass corridor and there no wifi networks at all. At this short moment I can see that caching process is stopped. So, in about two minutes I get to Building A. I sit down and keep watching the youtube clip. In about a minute browser continue to cache the clip. I check accessibility of other web-sites (including not being visited before) and they are OK. I sure, that at this moment I have connected to APs of Building A. BUT THEN HAPPENS THIS:

About in 5-10 minutes after getting to Building A clip caching stops and all the sites become inaccessible. Windows says: «Limited connection». I try to reconnect to my SSID and get the same result. Turning off/on wifi adapter doesn’t help. I try to restart the notebook – no results. I repeat the operation from B to A with my Nokia Lumia 930 – same things. A guy with Lenovo / Debian repeated it after me. I’ve waited more the 30 minutes and the wifi networks it still inaccessible. I found out, that I didn’t get an IP address and the only way to connect to wifi is to CHANGE MAC-address!!! Simultaneously, for all new clients dhcp and wifi work fine at Building A.

So, my question is: how to made roaming work as it should? Could I use wifi walking from one building to another without interruptions with my deployment?

Gentlemen, please, help! I could provide all the information required/you’ll request.

 

Many thanks in advance,

Ilya 






Photo of Ilya Semenov

Ilya Semenov

  • 4,384 Points 4k badge 2x thumb

Posted 2 years ago

  • 0
  • 1
Photo of Doug Hyde

Doug Hyde, Technical Support Manager

  • 20,192 Points 20k badge 2x thumb
If they are the same subnets on both A and B it should work fine. Few questions..

1. Have you checked time on the controller pairs, they need to match?
2. Are the controller configs synced between the pairs?
3. I know you said it was open but do you have any rules in your role assigned to the users?
4. Have you contacted GTAC to have them review the configuration files and station events on the controllers?
(Edited)
Photo of Ilya Semenov

Ilya Semenov

  • 4,384 Points 4k badge 2x thumb
Hello, Doug!

Thank you for your reply!

Yes, both controllers are at the same subnet.

1. I think the time is OK (its configured to use local NTP-server), but I'll check it on Monday.
2. How can I check whether the configs are synced? 
3. There are no any rules instead the default
4. The service contract for controllers have to be renewed in two months(((

Thank you.
Photo of Doug Hyde

Doug Hyde, Technical Support Manager

  • 20,192 Points 20k badge 2x thumb
VNS Tab-->Global (on left side)--> Sync Summary
Photo of Doug Hyde

Doug Hyde, Technical Support Manager

  • 20,192 Points 20k badge 2x thumb
Also, make sure the availability link is up. Reports-->APs--->AP Availability top of the page will display Availability Link is UP 
Photo of Ilya Semenov

Ilya Semenov

  • 4,384 Points 4k badge 2x thumb
OK, thanks, I will follow both recommendations on Monday. But I believe that 99% of APs are availaible all the time.
Photo of Doug Hyde

Doug Hyde, Technical Support Manager

  • 20,192 Points 20k badge 2x thumb
It's not to check the status of the ap's, it's more making sure the controller pairs are linked together. If that link is not formed, then you will experience the issue you noted above. Good luck on Monday and keep us posted. 
Photo of Ronald Dvorak

Ronald Dvorak, Embassador

  • 45,132 Points 20k badge 2x thumb
What is the topology that you use i.e. routed, bridge@AP, bridge@EWC

Is mobility enabled, GUI > Controller > Services > Mobility Manager
If yes is the Mobility Tunnel up - check on the dashboard (main GUI page) in the lower left corner - green&up or red&down.
Photo of Ilya Semenov

Ilya Semenov

  • 4,384 Points 4k badge 2x thumb
Hello, Ronald, hello,Doug!

Reporting:

1) I've checked the controllers configuration. There were great difference in time - about 5 hours - I've fixed it.
2) AP availaibility is OK - all the APs are green on both controllers.
3) The mobility were turned off. Now I turned if on on both controllers, one as manager and the second as client.
4) For a while, there are no mobility tunnels up.

Have I configured mobility in a right way?

Please, take a look at screenshots provided.

Many thanks in advance,

Ilya
Photo of Doug Hyde

Doug Hyde, Technical Support Manager

  • 20,192 Points 20k badge 2x thumb
You don't need mobility enabled if the same subnets are on both sides of the pair. The time being off should have fixed the original issue. 
Photo of Ilya Semenov

Ilya Semenov

  • 4,384 Points 4k badge 2x thumb
Doug, should I turn the mobility off?

You mean that correcting the time on both controller will solve the issue?

Thanks,

Ilya
Photo of Doug Hyde

Doug Hyde, Technical Support Manager

  • 20,192 Points 20k badge 2x thumb
Yes, I would turn mobility off. You don't need it on if the clients are in the same subnet on both pairs.  Time being off will cause the "stuck" issue if you roam the controller pairs. That's why it was the first questions I asked :-) 


1. Have you checked time on the controller pairs, they need to match?
Photo of Ilya Semenov

Ilya Semenov

  • 4,384 Points 4k badge 2x thumb
Yes, I've checked the time and now it's 16:38 on both controllers.

And I will turn the mobility off, as it was always...
Photo of Ilya Semenov

Ilya Semenov

  • 4,384 Points 4k badge 2x thumb
Ooops, I've forgotten to post the screens...
(Edited)
Photo of Ilya Semenov

Ilya Semenov

  • 4,384 Points 4k badge 2x thumb
And the last strange picture is from Netsigt - it doesn't show availability status...
(Edited)
Photo of Ronald Dvorak

Ronald Dvorak, Embassador

  • 45,132 Points 20k badge 2x thumb
Availability in Netsight has nothing to do with the controller availability pair - it's an indicator if Netsight was able to reach/poll the controller over time.




Has Netsight the user/right to access the contoller via ssh - this is needed to collect certain information.

Please check out this post...
https://community.extremenetworks.com/extreme/topics/using_netsight_5_to_manage_wireless_controllers
Photo of Ilya Semenov

Ilya Semenov

  • 4,384 Points 4k badge 2x thumb
Gentlemen, do I understand correct, that the problem with clients roaming between Building A and B was the great difference in time between two paired controllers?

Thank you.
Photo of Doug Hyde

Doug Hyde, Technical Support Manager

  • 20,192 Points 20k badge 2x thumb
Yes
Photo of Ilya Semenov

Ilya Semenov

  • 4,384 Points 4k badge 2x thumb
So, I'll try to check this solution tomorrow and report my results.

Thanks.
Photo of Ilya Semenov

Ilya Semenov

  • 4,384 Points 4k badge 2x thumb
Hi, Doug!

The reason for the issue was the great time difference between the controllers, you were right. After I've corrected it everything worked fine.

Many thanks for your advice, Doug!
Photo of Doug Hyde

Doug Hyde, Technical Support Manager

  • 20,192 Points 20k badge 2x thumb