So my setup is running on Amazon EC2 instance. Currently I have ALL ports being forwarded so essentially no firewall configured.
I can make and receive calls perfectly fine.
I noticed in testing this setup I had an issue that if a user called the IVR then the IVR rang a ring group it would ring all involved but the person that answered would hear nothing and neither would the other end. The calls didn’t disconnect just no audio. If I tried to call again every member of the ring group would ring except the people that have last tried to answer a call from that ring group. Almost as if Asterisk still thought they were on the call and didn’t ring the extension because of that. Long story short we don’t use ring groups so I never worried about it.
Today a user called AT&T, they were walking through their IVR and it went to transfer them to a support agent and as soon as AT&T initiated the transfer you hear to faint clicks and thats it. The call timer still runs so Asterisk didn’t end the call but I could no longer hear anything. Looking in the logs you see this.
-- Called SIP/BW-SIP-A/+18003310500
-- SIP/BW-SIP-A-00000317 is making progress passing it to SIP/120-00000316
-- SIP/BW-SIP-A-00000317 is making progress passing it to SIP/120-00000316
-- SIP/BW-SIP-A-00000317 answered SIP/120-00000316
[2013-02-14 17:01:25] NOTICE[1934]: chan_sip.c:25735 check_rtp_timeout: Disconnecting call 'SIP/BW-SIP-A-00000317' for lack of RTP activity in 31 seconds
[2013-02-14 17:01:26] NOTICE[1934]: chan_sip.c:25735 check_rtp_timeout: Disconnecting call 'SIP/120-00000316' for lack of RTP activity in 31 seconds
[2013-02-14 17:01:31] WARNING[1934]: chan_sip.c:3873 __sip_autodestruct: Autodestruct on dialog '4279fcaf-30ae2d0c-54cf9b69@10.0.2.5' with owner in place (Method: BYE)
[2013-02-14 17:01:58] WARNING[1934]: chan_sip.c:3873 __sip_autodestruct: Autodestruct on dialog '11c873e076a0e6627612f4fc09f84ded@107.22.237.xxx:5060' with owner in place (Method: BYE)
Which turns out to be the EXACT same errors I was getting months ago when testing the ring groups.
I use FreePBX 2.10 as a GUI and Bandwidth.com as my SIP provider.
SIP Settings has NAT set to yes, it has external IP proper and local networks as 10.0.0.0/255.0.0.0
Reinvite is set to Update and RTP is set to 30 seconds
I compiled Asterisk from source here is version output
Asterisk 1.8.12.0 built by root @ domU-12-31-39-06-2E-05 on a x86_64 running Linux on 2012-05-22 23:25:36 UTC
I am running Amazon Linux 64 bit 2012.03
Help with this would be greatly appreciated and please let me know if more info is needed.
Asterisk has no concept of IVRs or ring groups. These are constructs created by dialplans. You need to ask the person who wrote the dialplan, e.g. freepbx.org if you are using the FreePBX GUI.
Asterisk transfers completely remove the call from Asterisk, so it is unlikely that you are really doing a tarnsfer.
It looks like you have a NAT or firewall problem that is physically blocking the the RTP, but there is far too little information to say much more than that. You need to provide a trace showing the SDP exchange.
I don’t know enough about the EC2 environment, except that it is a virtual machine system, and therefore not designed for the tight real time constraints of VoIP, so likely to produce high jitter and choppy voice announcements.
We have been using asterisk on this EC2 instance now for well over 6 months without any issues of call quality etc. I am not sure if that is what you’re referencing or not by your mentioning of it however. Either way it works flawless except for this transfer issue. Everything I have researched says that RTP activity is a result of a firewall issue it just doesn’t make sense as a call that isn’t transferred works as expected. I would expect if it were a firewall issue this wouldn’t be the case, but that’s why I am posting on these forums though haha because I am at a loss.
You are not doing a transfer; you are setting up a tandem connection. The only way that would change the RTP handling is if you had directmedia=yes, but, if you had that, Asterisk would not be expecting to see any RTP, so wouldn’t be timing it out.
Even if your inline trace is only partial, and is actually the enquiry leg of an attended transfer, Asterisk still routes the call as though it had been a simple through connection. (I suppose in that case, you might just have had a system that worked because you were using directmedia, but are no longer using it. Direct media would also remove the effects of a VM on the through audio.))
If I have to work to get at a trace, I’ll normally not bother.
David55 thanks again for continuing to help me, I really appreciate it.
I am confused about your last statement in the reply, did you not see the link to the traces that I linked or are you saying they aren’t correct, I am confused.
