Time pattern for call drops

[lmora]

Hi all

My issue:
Calls being dropped/hung up in the middle of conversations

Information about my settings:

SIP trunks to dial out through ViciDial
eyeBeam softphones using g711 ulaw for conference
g729a for communication from ViciDial to Sip provider
4MB ADSL internet 3MB downstream / 426 kbps upstream
ViciDial server behind a firewall (Orange interface)
Ops computers in green interface
10 simultaneous calls at the most
2:1 ratio dialing

I've noticed that some calls are being dropped between 120 and 140 seconds of call duration without any error message of the event at neither CLI nor Firewall log. iftop also is not showing increased consumption of bandwith.

I still have to figure out if it is a correlated to CPU or query processing usage with mtop and htop commands. But honestly I don't believe the problem is there since my server is not small despite I have database and web server in the same box there are only 8 to 10 simultaneous calls at the most.

So I was wondering if these events could be because of some parameters in the configuration of the softphone.

could you give some optimal values for the parameters to be configured in a softphone?
should I send SIP keep-alives?
should I use rport?
should I allow SIP rport detection?


Thanks

[williamconley]

Try it with Two Simultaneous calls at once instead of 10. If you go a full hour with NO DROPS ... I expect that your .4M connection is "shared, best effort" DSL and not "continuous guaranteed" .4M.

Which can quickly cause this sort of issue.

That's why people pay for a T1 (1.5M) instead of just getting 3 .5M (512K) DSLs and a good multicasting router, the .5M isn't "real". You could also try using your router to REJECT all traffic other than SIP to your server. This would mean that the kid who's pulling your wifi from next door to occasionally watch a youtube video may be disappointed, but it may stabilize your net.