Francesco Castellano
2008-Apr-22 10:28 UTC
[asterisk-users] lots of warnings from translate.c
We have a couple of servers with asterisk 1.4.19 and zaptel 1.4.10, acting as gateways from SIP to ISDN PRI interfaces. Each has one Digium TE420 (with hardware echo cancellation) and one TC400B for transcoding, in order to handle 60/90 contemporary calls in peak hours. In my logs there are hundreds of thousand warnigs per day like these: transcode.c: no samples for lintoulaw transcode.c: zapg729toalaw did not update samples ### Could you suggest me what are the possible causes for that? Are they signs of bad audio quality? Any ideas for resolving these issues? In addition I can say that we are using a quite large jitter buffer in zapata.conf: jitterbuffers=16 (=> 0.32s) Moreover, it uses the fixed implementation, because when I tried the adaptive one I experienced one-way audio. Finally I have to note that, using a Siemens IP phone (G.729 no AnnexB) in conditions of no load on servers, I could replicate non-deterministically (sigh!) each of these problems, with a very noisy audio, and a annoying period of silence during the first seconds of call. Regards, Francesco PS. Previous versions of asterisk and zaptel presented an identical situation.
On Tue, 2008-04-22 at 12:28 +0200, Francesco Castellano wrote:> We have a couple of servers with asterisk 1.4.19 and zaptel 1.4.10, > acting as gateways from SIP to ISDN PRI interfaces. Each has one > Digium TE420 (with hardware echo cancellation) and one TC400B for > transcoding, in order to handle 60/90 contemporary calls in peak > hours. > > In my logs there are hundreds of thousand warnigs per day like these: > > transcode.c: no samples for lintoulaw > transcode.c: zapg729toalaw did not update samples ### > > Could you suggest me what are the possible causes for that? Are they > signs of bad audio quality? Any ideas for resolving these issues? > > In addition I can say that we are using a quite large jitter buffer in > zapata.conf: > > jitterbuffers=16 (=> 0.32s) > > Moreover, it uses the fixed implementation, because when I tried the > adaptive one I experienced one-way audio. > Finally I have to note that, using a Siemens IP phone (G.729 no > AnnexB) in conditions of no load on servers, I could replicate > non-deterministically (sigh!) each of these problems, with a very > noisy audio, and a annoying period of silence during the first seconds > of call. > > Regards, > Francesco > > PS. Previous versions of asterisk and zaptel presented an identical situation. >Have you tried additional types of phones and if so can you produce the same non-deterministic problems? Dave