Issue with packetization time between regions.(not sure how they are different)

Jay Schulze · ‎10-03-2017

Hello,

So ran into a problem that I haven't seen before. First I was trying to figure out why a transcoder was being used for these calls. Since both were using 711ulaw between the regions. Finally after going through traces I see this is the reason.

AFTER MATCHING LOGIC applied, capsA[1]::capCount=1 (Cap,ptime)= (4,30) capsB[13]::capCount=13 (Cap,ptime)= (4,60) (40,60) (41,60) (2,60) (5,60) (3,60) (10,60) (20,60) (18,60) (11,60) (12,60) (19,60) (9,60) numMatchedCaps=1

As you can see they advertise the right codecs but with different ptimes. The only area I know where this can be changed is the clusterwide parameter of 'preferred G.7XX Millisecond packet size. which are all default for 20. So where in the world is it getting that different ptime between the regions?

Jonathan Unger · ‎10-03-2017

Hi there,

Interesting issue indeed. A few questions:

Is the call between 2 local endpoints? Or an endpoint and a trunk/gateway?
What types/models are the 2 endpoints on the call (example 8841 phone and a 4331 SIP VGW)?
Are you messing with the audio codec preference list between the regions?
Have you tried resetting the devices you are testing with?

There is no other area that I am aware of (other than the parameter you mentioned) where you can adjust the ptime of CUCM registered endpoint calls. Additionally I have never heard such a thing as a 60ms ptime on G711u which is what I think "(Cap,ptime)= (4,60)" is implying support for.

It would be helpful if you could attach a full CM trace for the test call, along with the calling and called number.

Jay Schulze · ‎10-04-2017

Hello thanks for taking a look. It is unusual I have never seen a different ptime coming from cucm either.

1) The call flow is this. CVP--sip--CUCM--sip--LCP--sip--ICT--ip phone. So the devices this is happening between the regions are LCP and sip trunk. Now I was thinking maybe the packetization is different on the CUCM on the other side of the trunk. But you'll see from traces the caps mismatch comes even before the invite is even sent to the trunk.

2)described above with regards to endpoints.

3)the regions this is occurring to are between A=regMTP B=regCoimbatore. Attaching screenshot of codec preference list between the two. Yes it is different but only in that 711 is first.

4)yes

Not sure how to upload traces it won't allow me to upload .rar or .zip. And would be silly if having to upload all the sdl files.

Jonathan Unger · ‎10-04-2017

Seems like a not very helpful restriction to have on the new support forums...

Testing .ZIP upload by renaming to .txt.

Ok it should work if you zip them up and change the file extension from .zip to .txt. We can then download the file and change the extension back to .zip.

Jay Schulze · ‎10-05-2017

Yeah no idea why they did that. But that is good idea let me try it.