WLAN-phone 6120/6140 watchdog timeout

[WimNL]

We are running into Watchdog Timeout error messages inside a large 5 GHz WLAN using 600 wireless phones (type 6120/6140 including a 2245 cluster and CS1k system). The errors occur random a couple of times per day, even on a phone that was (not touched) on a desk directly underneath an Access Point. Voice quality and roaming work extremely well. SVP-support has been setup inside the WLAN (Trapeze MX's and MP-422 AP's). Wired phones work fine.

On Feb 4 MrHanMan reported simular problems (at this forum).

Has anyone else seen simular problems?

Michael McNamara posted a solution for Motorola WLAN for the same problem:

http://blog.michaelfmcnamara.com/2008/08/what-does-watchdog-timeout-mean-on-nortel-wireless-phones/

 

[DaddyOfThree]

If the phones are only rebooting while idle or displaying watchdog timeout your probably loosing packets somewhere in the network between the (Signaling Server -> 2245 -> 6120/6140)

The first thing you need to confirm is that your wireless infrastructure is VIEW certified by Spectralink/Polycom, which I believe Trapeze is just not sure of the specific version.

Do you have the WLAN (ESSID) setup for a DTIM of 3? Have you configured all switch ports for trusted QoS operation? Have you actually checked (packet trace from a port mirror) that the DiffServ QoS bits are being properly applied? At the Signaling server? At the 2245?

You need to start by checking all those things (and more). If everything checks out then you need to start your hunt for the needle in the haystack.

You'll need a few laptops/desktops to act as packet sniffers and you'll need to setup a wireless trace (this can be a challenge because the phone could jump to any channel in the 802.11a/b/g spectrum - so you need a packet sniffer per channel or you can try this product --

http://www.cacetech.com/products/airpcap-ex.htm)

and you'll need a trace of the 2245.

Wireshark can now decode UNIStim but it won't be able to decode the SVP protocol so you'll need the help of Spectralink/Polycom to decode the wireless traces.

The solution works it's just a that SVP and UNIStim are very chatty protocols and the timers are not very forgiving for a wireless environment where the phones using power save polling.

Good Luck!

 

[WimNL]

Thanks DaddyOfThree,

A couple of your tips were already checked, but some others might help us further. Polycom and Nortel are involved and have done some tests on-site yesterday and today. Their advise is to use only the lower channels on 5 GHz to avoid interference with other systems, like radar. We'll change this, to check if it will give any improvement. In the meanwhile we'll be glad to receive additional advises.

Many regards,
WimNL.

 

[edslopes]

http://blog.michaelfmcnamara.com/2008/08/what-does-watchdog-timeout-mean-on-nortel-wireless-phones/

 

[VoWLAN]

Hi WiMNL,


Did you find how to fix the problem.I have similar problem in one our site and I'm wondering how you could resolve this problem.

 

[WimNL]

Yes, we were able to tackle most of the problems. We had to work closely together with Nortel, Polycom and Trapeze to tackle multiple 'problems' in all three systems. The result was however very promissing thanks to the outstanding effort of all three vendors and the German caseleader of Nortel.

1. The CS1k needs an upgrade, timers are adjusted by Nortel for VoWLAN. There is an official release available now.
2. The SVP-servers have to run software release 36 or higher. Multiple '(potential) problems' are solved.
3. We do have a last issues with the WLAN, but the telephony installation runs OK.

Hope this helps.
Many regards,
WimNL.

 

[gulan]

Hi WinNL,
Thanks for the useful information sharing with all of us. We do have a deployment of 6140 with Nortel 2380 (Trapeze) wireless controller. We do had lot of issue and gone through lot of patches & upgrades. Finally it is improved a lot. Recently we have eliminate the "Watchdog Timeout" with signaling server patch, now we are still encountering "No SVP Response" even right under the AP. Nortel support team now advise to upgrade the Wireless Security switch to 7.0.5.6.

We are currently using .034 version of the software on the 2245 (SVP Server). I have checked on the Nortel website and could not find .036. I am just wondering how did you get the .036 version? What are issues resolved on .036 ? I am also curious to know what software version using on your 6140 ?

Also I appreciate if could share with us on what kind of issues that you are still having?

Thanks with regards
Gulan

 

[VinceWhirlwind]

There's an excellent explanation of the difference in troubleshooting Voice Quality and Service issues here:

http://www.voiptroubleshooter.com/diagnosis/usersymptoms.html

 

[Alexion]

Thanks daddyof3 we are using Cisco wlan and we have downloaded firmware into the 6140s but we are still getting no svp response is there a way to go round it or do we need to upgrade tftp server

 

[DaddyOfThree]

Are you using the latest and greatest firmware for the 6140?

We had a voice quality issue using the latest software release for our Motorola Wireless LAN switch until we upgraded to the latest release for the 6140, I believe it was firmware release 115.019.

You should really have a TFTP server setup somewhere in your network. It doesn't have to be a dedicated server, you could install it on almost any server that you have already running.

Cheers!

 

[Alexion]

we already have a tftp server setup with the latest firmware and the 6140s have already downloaded the firmware still no svp response

thanks daddyof3

 

[biv343]

You might want to look at patch MPLR27785 for the CS1000 signalling server. It made a big difference in the performance of the phones for several of my customers.

Below are the notes from ESPL:



Watchdog timeout and No PABX Responce *** NOTE: WIFI SETS (22xx and 61xx IP sets) SPECIFIC. SEE NOTES ***

WIFI sets (61xx/22xx) go periodically for 1.4 sec deep sleep mode and then
check RUDP messages for about 100 ms.
In case of changing AP registration could take a while.
So default RUDP settings 400msx10 times on CS1K side are not optimal.
Polycom proposal is to change timeout to 1600 ms and leave retries as is (10 times).

Notes

It's necessary to reboot wifi sets (61xx/22xx) after installation patch.

RUDP timeout for Wifi sets is set to 1600 ms (total = 1600ms * 10 retries).

To install this patch: Set Thresholds for patch to either 0, 100, 0 or 0,
0, 0. This will be the same as if the Deplist loaded this patch and prevent it
from being removed from service due to Thresholds being reached.