Kristian Kielhofner
2005-Mar-19 15:24 UTC
[Asterisk-Users] More HEAD wierdness (chan_sip, jitterbuffer/PLC problems)
Hello, After checking out CVS HEAD from yesterday (for those new PLC/Jitterbuffer patches), I was affected by bug 3795 with my Polycom IP600's. After seing it resolved as of this morning (thanks Mark), I decided to try again... I can answer incoming calls. No problem there. Putting calls on hold, however, results in my Polycom IP600 indicating the call on hold, but the caller does not hear any music. If I go through enough cycles of holding/resuming the call, the IP600 eventually locks up in hold, with the resume softkey doing nothing. The only way to get out is to have the caller hangup or reset the phone. If the caller hangs up, the Polycom drops the call, and a few seconds later the caller's phone rings as though the Polycom was calling it back. If you answer the call you hear nothing and the call is dropped after about one second. Similar behavior on my 7960. I can place the call on hold, but the caller hears no music. Same with before, if I cycle hold/resume enough, the phone locks up. Only the Cisco just drops the call. The SIP debug of the conversation between * and the IP 600 (firmware 1.4.1) can be found below. I can provide the Cisco as well (7.2). http://www.krisk.org/asterisk/sip_debug.log Now for more on MOH. I am using native MOH with my files in ulaw format. I setup an extension for running MusicOnHold, and have been listening to my hold music for 8 minutes with my Polycom IP600, so I know that * sees the files and can play them. However, if someone calls in over IAX and executes that same extension, MOH plays for a couple of seconds and stops. No word on the * console as to what happened. I use g729 almost exclusively. I am confused as to the status of g729 with the new PLC/jitterbuffer stuff, but that doesn't seem to be it because I can transcode from ulaw -> g729 on the phone with my MOH extension with no problems. "sip show channels" shows that the channel is using g729, and the MOH plays for forever and ever. My iax.conf has trunk=no trunktimestamps=no and jitterbuffer=yes. If I set jitterbuffer=no, MOH works perfectly. So it appears to be the jitterbuffer in IAX that is causing problems with MOH there. I hope that this message wasn't too verbose, but I hope that any of these PLC/jitterbuffer problems can get ironed out, because it looks awesome!!! -- Kristian Kielhofner
Kevin P. Fleming
2005-Mar-19 17:05 UTC
[Asterisk-Users] More HEAD wierdness (chan_sip, jitterbuffer/PLC problems)
Kristian Kielhofner wrote:> After checking out CVS HEAD from yesterday (for those new > PLC/Jitterbuffer patches), I was affected by bug 3795 with my Polycom > IP600's. After seing it resolved as of this morning (thanks Mark), I > decided to try again...There have been more chan_sip fixes committed today (including one for bug 3798 just a few minutes ago), so please update your system and try again to see if the problems persist. Thanks for testing!
Kristian Kielhofner
2005-Mar-19 18:44 UTC
[Asterisk-Users] More HEAD wierdness (chan_sip, jitterbuffer/PLC problems) -UPDATE
Kristian Kielhofner wrote:> Hello, > > After checking out CVS HEAD from yesterday (for those new > PLC/Jitterbuffer patches), I was affected by bug 3795 with my Polycom > IP600's. After seing it resolved as of this morning (thanks Mark), I > decided to try again... > > I can answer incoming calls. No problem there. Putting calls on > hold, however, results in my Polycom IP600 indicating the call on hold, > but the caller does not hear any music. If I go through enough cycles > of holding/resuming the call, the IP600 eventually locks up in hold, > with the resume softkey doing nothing. The only way to get out is to > have the caller hangup or reset the phone. If the caller hangs up, the > Polycom drops the call, and a few seconds later the caller's phone rings > as though the Polycom was calling it back. If you answer the call you > hear nothing and the call is dropped after about one second. > > Similar behavior on my 7960. I can place the call on hold, but the > caller hears no music. Same with before, if I cycle hold/resume enough, > the phone locks up. Only the Cisco just drops the call. > > The SIP debug of the conversation between * and the IP 600 (firmware > 1.4.1) can be found below. I can provide the Cisco as well (7.2). > > http://www.krisk.org/asterisk/sip_debug.logFixed in CVS. Thanks again.> Now for more on MOH. I am using native MOH with my files in ulaw > format. I setup an extension for running MusicOnHold, and have been > listening to my hold music for 8 minutes with my Polycom IP600, so I > know that * sees the files and can play them. However, if someone calls > in over IAX and executes that same extension, MOH plays for a couple of > seconds and stops. No word on the * console as to what happened. > > I use g729 almost exclusively. I am confused as to the status of > g729 with the new PLC/jitterbuffer stuff, but that doesn't seem to be it > because I can transcode from ulaw -> g729 on the phone with my MOH > extension with no problems. "sip show channels" shows that the channel > is using g729, and the MOH plays for forever and ever. My iax.conf has > trunk=no trunktimestamps=no and jitterbuffer=yes. If I set > jitterbuffer=no, MOH works perfectly. So it appears to be the > jitterbuffer in IAX that is causing problems with MOH there. > > I hope that this message wasn't too verbose, but I hope that any of > these PLC/jitterbuffer problems can get ironed out, because it looks > awesome!!!Sorry to reply to my own post, but here is a new interesting tidbit of information: If a call comes in and dials my MOH extension, it plays MOH for a few seconds before stopping. If a call comes in, dials a SIP extension and answers the call, the audio is great (two humans talking). If I place that call on hold, the MOH sounds fine too (for the 2 minutes that I listened). Truely strange. I don't think I mentioned that the other end of that IAX2 connection is NuFone. But I have tried SIxTel (iax.cc) and it has the same result. -- Kristian Kielhofner