[asterisk-users] oslec using sample.c for long(er) dumps

Leon de Rooij leon at scarlet-internet.nl
Thu Mar 5 09:10:36 CST 2009


Hello all,

Since a while some of our SIP users complain about gaps (sometimes  
multiple seconds of silence) in the RX audio stream (direction pbx ->  
phones). Our configuration is an Asterisk with two Wildcard TE410P  
cards that are connected with E1 PRI's to an external server running  
callcenter software. The SIP users use Snom and Polycom phones. The  
gaps are not present in recordings on the external callcenter server.

(there are no problems with calls going through the pbx - where rtp  
also goes through the pbx - that don't use the PRI's)

I tcpdumped all traffic on various points in the network between the  
phones and the pbx and noticed that the gaps are already present on  
the ethernet interface of the pbx, so I ruled out network problems.
(It's not packetloss, just silence - I converted the pcap files to  
audio files and listened to it)

Next step is to see whether the gaps originate on the Asterisk pbx or  
the external machine, so I wanted to dump traffic directly from the  
zaptel interface.

The supplier of our pbx suggested that it may be possible to use a  
feature of the oslec patch for zaptel (see below), which has a sample  
utility that can dump a channel to disk. I've been looking through  
sample.c, but I see there is a maximum sample time at line 97, though  
I don't see why this maximum exists. The buffers seem to be written to  
disk every iteration of the for loop, so it shouldn't fill up memory  
(correct me if I'm wrong).

Does anyone have any tips or suggestions, whether I can just remove  
this limit ? Or is there a better way to dump zaptel traffic for later  
inspection ?

Thanks and kind regards,

Leon de Rooij
leon at scarlet-internet.nl




( http://www.rowetel.com/ucasterisk/oslec.html#sample )
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.digium.com/pipermail/asterisk-users/attachments/20090305/33366468/attachment-0001.htm 


More information about the asterisk-users mailing list