1 way speach path issue on CSE1000

[VOIPaintEASY]

We have multiple CSE1000's at different locations. 1 as large as 600 IP sets, others as small as 20, 50, 70 IP sets. On all of these systems we are at various levels of release. They all share an occasional problem that seems to be a widespread CSE1000 issue. It is one way speech path. What I mean is a call will come in or go out the PSTN, across virtual IP trunks, set to set, etc and only 1/2 of the conversation can be heard. If you redial it likely will connect correctly. We have gone for over a year with this problem building at all locations. I think it is not reported because folks can redial and the next call will work. With the reduced expectation of cell calls in the past 10 years I think people have increased their tolerance to unpredictable call establishment. Well, after over a year folks are coming to the conclusion this is a system issue and reports are near daily from all systems. Anyone else with a CSE seeing this issue?

 

[dlesap]

How your PBX are connected?
Signalling server?
ITG?

Do you have a NRS or Gatekeeper?

Which releases for those PBX? All the same?

 

[VOIPaintEASY]

We have signalling server in all instalations. They are pure CSE1000's, not Option X with an ITG card.

 

[dlesap]

OK. The way to debug that problem is to connect with telnet on Sigserv and to trace the calls. Log in with admin1 level user and pwd.

Make a call and you may received info. If not, you have to activate trace. Let me know if OK or not, cause i don't remember by heart the command to activate.

Infos you will receive are the ISDN/H323 signalling negociation. That's including codec messages where i guess you have a problem.

 

[VOIPaintEASY]

dlesap,

Have you seen this specific issue and the fix being codec related?

 

[dlesap]

Clearly not. Be carefull all patches are up to date on CS, SS and MC. If possible, try to remove G729 codec in all SS and try again. Sometimes, that kind of fault is a codec negociation problem. If you have only one codec, no negociation problem...

Are you able to replic the problem in a specific case? If so, you may isolated the problem by checking which equipments are involved.

I've also seen some sporadics faults on Media cards when DSP are used. Reboot MC is a possible solution.

 

[SysTech2]

Make sure your elan connections are all connected to a switch not a hub. Do you have all your IP phones on a seperate vlan, isolated from data. Did you check all your switch configs?

 

[VOIPaintEASY]

Elan is on a switch. We have separate voice and data VLAN/s for each floor. We also run OPS for QoS policy. If it was a QoS issue it would not be totally dead we would have echo or jitter (been down that path too). It is a true 1 way only speach. 1 way is totally dead.

 

[dlesap]

You have to sniff the concerned call. 1 sniff behind the set, 1 sniff behind SigServ TLAN, 1 sniff behind Media Card TLAN.In that traces you may find UDP messages with port number: port 5000 for one way, port 5100 for the other way. If any of these messages are not present, try to see where it disappear in the Network.