Well, At first, I was just learning.... Then, it became a troubleshooting process. I have box1 acting as a PSTN gateway with a TE405P. Box2 is used for our call center agents to login,answer,make calls and where calls are all recorded. I've been trying to troubleshoot a call-drop problem I continue having and have had for months. It is becoming seriously irritating and demoralizing to the point that upper management is almost decided to drop the whole asterisk-based project. As a test, I divided my agents in two groups. One group will take/send calls to box1 using IAX2 and the other using SIP. Well, for the past 2 days, I've noticed that the number of call drops in the group using SIP to connect the two servers has SIGNIFICANTLY reduced. For example, the group using IAX2 suffers from about 80-100 call drops a day, whereas the group using SIP suffers from about 20-30 call drops. This is from a sample of about 2000 calls per day. It may or may not seem much, but is is a significant number of call drops. Even 20-30 is not acceptable, but it's improving. I have asked for support from the list in the past regarding this issue but have not received much help. I even purchased a TDM400P board to put in box2 as a timer source to replace ztdummy but that hasn't helped. I've done network troubleshooting as much as I can to no avail. All network interfaces show 0 collisions, 0 error packets, 0 CRC errors, 0 packet loss. I have done random ping floods when agents experience call drops but everything turns out just fine. All agents are using Windows-based PCs with eyeBeam as softphone. The only "odd" thing I have seen is that when I run: ping -i 0.1 -c 10000 {ip_address} in order to almost flood ping an agent's station from the box2 asterisk server, the average latency is .2ms but every 1000 packets (approx), the latency jumps to 1ms. I would assume this would, at worst, affect call quality (jitter) but not a call drop. From a post I read this weekend, someone mentioned that maybe one (or more) of the PCs could be infected with a virus which could be causing this. I've been tempted to format a few machines with Linux and X-Lite to see if that improves. BTW, I've tried using IAX-based softphones and the experience is still similar with regards to the call drops. box2 is a Dual-Xeon 2.4GHz, 2GB RAM, 3ware SATA RAID 5, using the onboard Intel 10/100/1000 Etherexpress NIC, and handles merely 25 agents. I manually configured the interrupts to make sure CPU1 handles all TDM400P interrupts while CPU0 handles the rest (as some of the documentation suggested) but problem still persists. I don't do anything fancy in Asterisk. Answer a call, put the call on queue, deliver to agents using rrmemory strategy. That's all. I was even thinking of getting an account with a VoIP provider in order to divert some traffic there and see if asterisk connecting directly with a VoIP provider will handle my setup better. From what I've read on the list, I have a very very very simple setup and it doesn't make any sense for me to be getting this many call drops. I'm simply running out of ideas and almost ready to throw in the towel. I hope this has answered your curiosity and hopefully will allow you or others in helping me troubleshoot this issue further. Thanks, Waldo On Dec 6, 2005, at 10:22 PM, Alvaro Parres wrote:
|
_______________________________________________ --Bandwidth and Colocation provided by Easynews.com --
Asterisk-Users mailing list To UNSUBSCRIBE or update options visit: http://lists.digium.com/mailman/listinfo/asterisk-users