Call Transfer Problem

Unanswered Question
Oct 1st, 2009

HI

Anybody knows how work the call transfer leg between CallManager server on Cluster over wan (5.1)

We have a cluster over wan with 4 server, site A(central, 2), B(1) and C(1).

So when a phone (x) on site B call phone (y) on site C, the call works fine, but when phone y transfer the call the transfer had delay about 3 seconds.

When phone (y) on site C call phone (x) on site C, the call works fine and when phone y transfer call, don't have delay.

Thanks

Peterson

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
William Bell Thu, 10/01/2009 - 12:00

Peterson,

In your first scenario:

phone(x) on site(b) to phone(y) on site(c)

Is the delay occurring when phone(y) calls the transfer destination? or when phone (y) hits transfer softkey a second time (which actually transfers phone(x))?

Is phone(x) registered to CM in site(b)?

Is phone(y) registered to CM in site(c)?

Is there a delay when phone(x) calls phone(y)?

What is your RTT between:

phone(x)site(b) to CUCM at site(c)?

CUCM at site(b) and CUCM at site(c)?

Is the transfer destination in your scenario at site(c)? or another site?

Have you tested phone(x) registered to CUCM at site(c)? Thus localizing the issue further to either phone-to-cucm delay or cucm-to-cucm delay.

Remember in CUCM 5.1, your ICCS delay budget is 40ms.

pgcristovam Thu, 10/01/2009 - 12:41

phone(x) on site(b) to phone(y) on site(c)

Is the delay occurring when phone(y) calls the transfer destination? or when phone (y) hits transfer softkey a second time (which actually transfers phone(x))?

Second time.

Is phone(x) registered to CM in site(b)?

Yes

Is phone(y) registered to CM in site(c)?

Yes

Is there a delay when phone(x) calls phone(y)?

Where can I see?

phone(x)site(b) to CUCM at site(c)?

How can i see?

CUCM at site(b) and CUCM at site(c)?

between 40 and 50 ms

Is the transfer destination in your scenario at site(c)? or another site?

site c

Have you tested phone(x) registered to CUCM at site(c)? Thus localizing the issue further to either phone-to-cucm delay or cucm-to-cucm delay.

I will test.

Thanks

Peterson

William Bell Thu, 10/01/2009 - 12:49

With respect to:

>Is there a delay when phone(x) calls phone(y)?

>>Where can I see?

The issue sounds like you are having a delay with audio cut through. So, your user's would report a delay after the first call is answered. So, a noticeable clip in voice stream when the call starts.

The 40-50ms latency raises an eyebrow. Could be a contributing factor.

Is this a hub-spoke topology with a (b)-(a)-(c) network path?

What happens when you use a phone at site (a) to call site (c) and do the transfer action?

pgcristovam Fri, 10/02/2009 - 04:14

Hi

Don't have delay on call... rtp stream works fine. The delay is just on second transfer sofkey press.

Now, this topology is hup-spoke. Before was full-mesh and all system works fine. But now, is not possible the topology is hup-spoke (policy of customer).

Delay from telephony x on site B to CUCM on site C

same time the AVG is 44.829 ms

same time the AVG is 46.382 ms

same time the MAX is 52.401 ms

Delay from telephony y on site C to CUCM on site B

same time the AVG is 47.099 ms

same time the AVG is 51,439 ms

same time the MAX is 70.863 ms

pgcristovam Fri, 10/02/2009 - 05:56

Hi

But, my main question is if on both scenarios have some difference between transfer signalization.

I can't see it. And if I can't see difference, I can't see the reason of problem.

I know that on transfer between phone on different CUCM.

Thanks

Peterson

William Bell Fri, 10/02/2009 - 06:52

Your average RTT is outside of spec on CUCM 5.1. The maximum one way delay is 20ms (or 40ms round trip). You said that things worked fine when full-mesh. Now you are hub-spoke. If this is the only change in the network, it is good indication that the network latency is playing a role.

Is there something that leads you to believe that something else is the problem?

Have you tried have all phones involved in the transaction registered to the same CUCM? Is there a difference in behavior?

If you want to dive deeper to find which subprocess is introducing or impacted by the latency, then you will need to capture detail traces from a test call. I'd also grab sniffer traces is you can. If you grab sniffer traces from all parties, then you can construct the time line between transactions.

A 3 second delay will stand out like a sore thumb.

pgcristovam Fri, 10/02/2009 - 09:11

Is there something that leads you to believe that something else is the problem?

No, I really can't see why on scenario works and another no.

Have you tried have all phones involved in the transaction registered to the same CUCM? Is there a difference in behavior?

No, the transfer also don't works.

Thanks

Peterson

Actions

This Discussion