Hi everyone we are using kannel 1.4.4 and making some tunning to our plateforme wich has the following configuration:
1- kannel server 4vcpu 8g ram with max httpd request set to 1000 2- A webportal with 8 vcpu 32 g ram with max httpd request set to 5000 3- A Postgresql server 8 vcpu 32 g ram with max pgsql pross set to 5000 too i dont want to enter in details tunning those server (sysclt.conf, limits.conf ...) but they are well done assuming we know how to do it from our web servers, we are testing our sms plateform with fakesms, and we got a bottleneck that we want to understand and tune it up if we can launching the follwing test : /usr/local/kannel/test/fakesmsc -i 0.001 -m 20000 "216xxxxxxx xxxxx xxxx xxxx 111111*11111111" wich we think means sending 20000 per 0.001 s at some point, when the launched commands ends, with fakesmsc command reaching the 20000 line we just got arround 6000 sms stored on the database but the flow drops down drastically without reaching the apache limit on kannel neither on the portal 800 httpd process not evenclose to 5000 wich can handle, and the stored sms continue with too slow processing, than looking at kannel status on port 13000 the store size about 14000 and the smsbox connections about the same ?! just by looking at that we just wonder if the bottle neck is there in the process of passing messages between smsbox and the bearerbox, and if so is there a way to tune that up if the system can handle it think you for any help