Okay, this is helpful. So you're definitely seeing RFC2833 digits from your telephone device. So the next question is what is going out to your provider. It just so happens that you didn't capture that leg of the call, but that's okay, you can just make another test call. Capture the packets heading out to Flowroute and see what's in there. I would listen to that capture in the Wireshark player and see if there are any digits going out.

Also, yes, the multiple packets for a single RFC2833 digit is quite normal.

-MC


On Tue, May 21, 2013 at 1:43 PM, Steven Schoch <schoch+freeswitch.org@xwin32.com> wrote:
I had the opportunity to capture the data to/from the phone. I used this packet capture command, on the FS host itself:

# /usr/sbin/tcpdump -p -s 0 -w jon.pcap -i eth1 host 192.168.4.235
10169 packets captured
10169 packets received by filter
0 packets dropped by kernel

The number dialed was 18002729872, the IRS EFTPS IVR.  The first voice prompt is to enter the 9-digit taxpayer identification number.

In Wireshark, I can clearly see the numbers being entered, as RFC2833 RTP DTMF events. One thing I notice is that I get several RTP packets for each digit. (I assume that's normal, but I'm new at this.) For example, here is one digit:

6507 87.427666 192.168.4.235 192.168.4.1 RTP EVENT 60 Payload type=RTP Event, DTMF Two 2
6508 87.431287 192.168.4.1 192.168.4.235 RTP 214 PT=ITU-T G.711 PCMU, SSRC=0x71C6FAAF, Seq=14414, Time=499365981 
6509 87.447662 192.168.4.235 192.168.4.1 RTP EVENT 60 Payload type=RTP Event, DTMF Two 2
6510 87.451285 192.168.4.1 192.168.4.235 RTP 214 PT=ITU-T G.711 PCMU, SSRC=0x71C6FAAF, Seq=14415, Time=499366141 
6511 87.467655 192.168.4.235 192.168.4.1 RTP EVENT 60 Payload type=RTP Event, DTMF Two 2
6512 87.471284 192.168.4.1 192.168.4.235 RTP 214 PT=ITU-T G.711 PCMU, SSRC=0x71C6FAAF, Seq=14416, Time=499366301 
6513 87.487656 192.168.4.235 192.168.4.1 RTP EVENT 60 Payload type=RTP Event, DTMF Two 2
6514 87.491286 192.168.4.1 192.168.4.235 RTP 214 PT=ITU-T G.711 PCMU, SSRC=0x71C6FAAF, Seq=14417, Time=499366461 
6515 87.507665 192.168.4.235 192.168.4.1 RTP EVENT 60 Payload type=RTP Event, DTMF Two 2
6516 87.511285 192.168.4.1 192.168.4.235 RTP 214 PT=ITU-T G.711 PCMU, SSRC=0x71C6FAAF, Seq=14418, Time=499366621 
6517 87.527677 192.168.4.235 192.168.4.1 RTP EVENT 60 Payload type=RTP Event, DTMF Two 2 (end)
6518 87.531284 192.168.4.1 192.168.4.235 RTP 214 PT=ITU-T G.711 PCMU, SSRC=0x71C6FAAF, Seq=14419, Time=499366781 
6519 87.547688 192.168.4.235 192.168.4.1 RTP EVENT 60 Payload type=RTP Event, DTMF Two 2 (end)
6520 87.551336 192.168.4.1 192.168.4.235 RTP 214 PT=ITU-T G.711 PCMU, SSRC=0x71C6FAAF, Seq=14420, Time=499366941 
6521 87.567658 192.168.4.235 192.168.4.1 RTP EVENT 60 Payload type=RTP Event, DTMF Two 2 (end)

This sequence happens 9 times with different numbers, as the 9-digit TIN is entered. At 80.5 seconds (according to the Wireshark player), the called party says: "The tax identification number must be nine digits. Please re-enter." The last digit event was at timestamp 2495866430 (phone) or 499404541 (FS). I don't know how to convert RTP timestamps to actual time.

What's the next step to diagnosing the problem?

-- 
Steve


On Thu, May 9, 2013 at 1:14 PM, Michael Collins <msc@freeswitch.org> wrote:
Are you sending in-band DTMFs or RFC2833? Or both? Get a pcap of the call, including the RTP, and analyze it in Wireshark. See what you are actually sending. That means looking at the flow (to see if you send RTPEVENTS, i.e. RFC2833 DTMFs) and also listening to the decoded audio with the player.

-MC


On Thu, May 9, 2013 at 12:59 PM, Steven Schoch <schoch+freeswitch.org@xwin32.com> wrote:
One of my users complains of many problems calling automated systems that take DTMF input. In this case it was an IRS payment entry system that would skip or duplicate button presses.

Our setup is pretty generic:
Polycom SoundPoint IP 320 SIP -> dedicated LAN -> Freeswitch (on CentOS, on XenServer) -> Comcast static IP -> Flowroute.com -> some IRS 800 number.

Is this just the way it is, or is there something I can do to make it work better?

-- 
Steve



--
Michael S Collins
Twitter: @mercutioviz
http://www.FreeSWITCH.org
http://www.ClueCon.com
http://www.OSTAG.org