VicidialNOW works slow after few hours

[erico1981]

hello,

i have installed VicidialNOW on my server ( 2GB of ram, 3.0GB XEON CPU,.. ) When i upload a new campaign, the first couple of hours works fine. all 30 agents are making calls out, but after that it starts dialing really slow! like 6 calls out of 30, so then i have to reboot the server so it will work fast again

memory usage while dialing is 40% of the 2GB only, why does it get so slow if not a memory issue?

P.S. i`m using g729 codec, SIP protocol. 5MB upload rate on the server.

thanks

[okli]

What is loadavg when this happens?
What type of XEON CPU do you have? Is it the old XEONs based on Pentium 4 or the new multi-cores? Is it a single CPU?
30 agents with g729 codec would require powerful machine if this is single server setup as it appears...

Better off post the results of the following commands:

Code: Select all

cat /proc/cpuinfo
cat /proc/loadavg

Issue the latter command when dialer has slowed down.

You may also monitor with 'top' if there is a process which consumes lots of CPU resources, and with 'mtop' if there is a slow SQL query.
Is this a new install or it has been working for a while?
What kind of outbound dialing are you using? If it is predictive- what's the dial ratio when this happens? What is drop percentage then?

[erico1981]

hello and thanks for ur reply, i will try to answer most of your questions

cat /proc/loadavg
----------------------
1.04 1.07 1.05 1/193 26160


cat /proc/cpuinfo
--------------------
model name : Intel(R) Xeon(R) CPU 3065 @ 2.33GHz
stepping : 11
cpu MHz : 2327.633
cache size : 4096 KB

cat /proc/meminfo
---------------------
MemTotal: 2074068 kB
MemFree: 1362684 kB
Buffers: 27804 kB
Cached: 501860 kB


this install has been working for like 4 months, i have the Dial method in the campaign set on RATIO and the AUTO DIAL LEVEL set on 1.

the cpu has 2 cores ( dual core ) as shown in the cpuinfo.

[okli]

Anything interesting in asterisk CLI when this happens? Can you post some output?

You may also monitor with 'top' if there is a process which consumes lots of CPU resources, and with 'mtop' if there is a slow SQL query.