[asterisk-users] oddity with TDM400P / Asterisk setup

Rich Adamson radamson at routers.com
Thu Aug 31 06:26:46 MST 2006


Ted Wallingford wrote:
> Hi List,
> 
> I am working with an Asterisk server running on Fedora Core 4. It has 
> two TDM400P cards installed. There are 6 trunk ports and 2 (unused) 
> analog line ports.  There are 5 Polycom SoundPoint 501 SIP phones 
> connected to the server, and a Linksys 24-port powered switch connecting 
> everything.  The * version running is 1.2.7.1.   All of the ports on the 
> switch with voice devices, including the server, have a service class of 
> 5, while non-voice devices are connected to other ports that have a 
> service class of best effort.
> 
> The problem, which began this morning, is very elusive.  
> Calls-in-progress from zap-to-sip or sip-to-zap or sip-to-Asterisk will 
> drop at odd times during the call, anywhere from 2 minutes to 15 minutes 
> into the call.   At the same time the call drops, my SSH session to the 
> server will hang. After 10 to 15 seconds, the output and input from ssh 
> session appears on my terminal and I am able to resume working in the 
> shell.  Zap-to-Asterisk doens't seem to cause the problem. Only when I 
> dial through to a SIP device does it seem to hang.
> 
> Top reveals nothing out the ordinary, utilization wise, the disk has 
> plenty of free space, and the arp cache doesn't ever indicate a 
> duplicate IP address with the server's NIC, which I thought might have 
> been the problem.  I also attempted to move the server to another port 
> on the switch. No improvement.  
> 
> Anybody have a problem like this?

Have not seen anything close to that problem.

You might check the linksys switch to see if it has Spanning Tree turned 
on. Spanning Tree (depending on vendor code) will disable a port from 
forwarding traffic for about 10 to 15 seconds as a means of detecting 
layer two loops. If it is turned on, turn it off and test again.

Also, you should be able to set up a series of pings from different 
sources to determine exactly which component in the infrastructure is 
failing.




More information about the asterisk-users mailing list