| Summary: | ASTERISK-23145: Sporadic one way audio between a SIP hard phone and a SIPML5 browser client on LAN | ||||
| Reporter: | Rusty Newton (rnewton) | Labels: | |||
| Date Opened: | 2014-01-15 19:11:22.000-0600 | Date Closed: | 2017-12-18 10:56:20.000-0600 | ||
| Priority: | Critical | Regression? | |||
| Status: | Closed/Complete | Components: | Channels/chan_sip/General Channels/chan_sip/WebSocket Resources/res_rtp_asterisk | ||
| Versions: | SVN 12.0.0 | Frequency of Occurrence | Constant | ||
| Related Issues: |
| ||||
| Environment: | Asterisk SVN-branch-12-r405587 Chrome Browser Version 32.0.1700.77 SIPML5 demo http://sipml5.org/call.htm?svn=203 | Attachments: | ( 0) extensions.txt ( 1) full.txt ( 2) http.txt ( 3) messages.txt ( 4) rtp_fail.pcap ( 5) rtp.txt ( 6) sip_debug_jssip.txt ( 7) sip.txt ( 8) sipml5_config1.png ( 9) sipml5_config2.png | ||
| Description: | Ran across one way audio issues that I believe stem from a bug while testing calls between two SIP clients (6001 and sipml5_chrome in sip.conf) in the process of building some documentation.
h3. Endpoints in sip.conf *6001* is a Digium D40 *sipml5_chrome* is SIPML5 on the sipml5.org demo running on Chrome. h3. Problem description * When *calling from 6001 to sipml5_chrome*, approximately one out of every three to six calls has one way audio, with audio flowing from 6001 to sipml5_chrome, but not the other way around. * When *calling from sipml5_chrome to 6001*, I appeared to get two way audio every time. h3. Environment SIPML5 is running on a Chrome browser, on the machine running Asterisk. The Digium phone is on the local LAN. There is no hardware or software firewalls or NAT going on. h3. Files attached h5. Config files extensions.txt http.txt rtp.txt sip.txt h5. Asterisk logs full.txt messages.txt h5. Packet capture rtp_fail.pcap The pcap contains several calls. Out of the six calls at the end, the first five had one-way audio, and *the very last call had two-way audio* (should be the very last call in the pcap. h5. SIPML5 configuration sipml5_config1.png sipml5_config2.png | ||||
| Comments: | By: Andrew Nagy (tm1000) 2014-01-24 22:12:12.235-0600 I've got the same sporadic issues with WebRTC going from jssip <--> UDP Endpoint (mainly outbound trunks) as Rusty has but in Asterisk 11.7.0 (note I am also using jssip, if I switch to sipml5 the results are nearly the same). At Matt Jordan's request I have uploaded the SIP Debug of my call session. (this was over a VPN which is why I haven't removed the IP addresses), If I dialed the internal echo test or any extension directly connected to Asterisk I have 2 way audio in 99% of tests, but when I attempt to dial an outbound number or if I do an inbound call from an external source I am plagued with the same issues as Rusty. By: Andrew Nagy (tm1000) 2014-01-24 22:15:08.822-0600 SIP Debug on outbound call while using jssip in the browser By: Matt Jordan (mjordan) 2014-02-27 16:52:39.573-0600 I linked this over to ASTERISK-22911 mostly because the patch Jonathan is working on resolves a fair number of one way audio issues, which may include this one. You may want to take a look at r409129/409130 and https://reviewboard.asterisk.org/r/3275 By: Jonathan Rose (jrose) 2014-02-28 14:58:17.922-0600 I don't believe my patch will fix the problem described as I've been running into it as well and I'm not aware of any fix for it just yet. I currently theorize that it may have something to do with the Controlled/controlling parties of the ICE session described in this issue: ASTERISK-23026 EDIT: Actually, I don't seem to be running into this problem anymore at the start of calls from my desk phones to SIPML5 clients. I'm just getting stuck in SRTP unprotect failure loops when holding/unholding them from SIPML5 when SIPML5 was the called party. So it's worth a shot to test this after all. By: Joshua C. Colp (jcolp) 2017-12-18 10:56:20.823-0600 Suspending as this hasn't been seen in quite a long time and things have changed. | ||||
