We have a cluster enviroment setup consisting of
2 x webservers (not load balanced but heartbeat if it needs to failover) - 4GB RAM Dual Core VM's
3 x Asterisk Nodes - 4GB Quad Core Servers
1 x DB - 32GB Dual Quad Core with mysql.sock etc mounted to a SAN (with a 2nd DB heartbeated and replicated for failover)
1 x Stats Server 8GB Quad Core VM's (replicated from main DB)
1 x IIS Webserver for scripts
with have no load issues on any of our servers. The webserver stays around 0-02 - 2.0, asterisk nodes around 28% - 34%, DB 0.05 - 1.8, IIS 1% - 4%
with have 51 agents logged into 5 campaigns across vicidial at peak times. ALL FORCE recording on each campaign. If you are wondering why we have a IIS server its because we have the script pop from the IIS server which has a custom dll file that pulls the lead_id from vicidial agent screen and populates the script. This has been working without any delay in the script for 3 months now, but has started to lag abit over the last few days.
VERSION: 2.4-362a
BUILD: 120316-1203
"Maybe" Other infomation you might need
- The mix recording perl script still runs throughout the day
all calls get transferred over night at 1am to external ftp
we have not turned on the archive_log script yet (vicidial_log = 755k rows, call_log=1.3m rows, vicidial_agent_log=570k rows, vicidial_carrier_log=766k rows, vicidial_log_extended=756k rows)
2 campaigns dial predictive max level 2.0 rest on ratio level 1.0
Like i said this has been working fine for 3 months but now has started to lag. When the call comes through to the agent interface the script_panel comes up white and then after 2-10 seconds sometimes longer the script will come on and populate the details.
I have checked the logs on the webservers/asterisknodes/DB/script server and moved the old logs off the servers. I have also checked the DB locks/peaks and its stable throughout the day with only 1-2 locks max.
Hope you can help with this and if you need any more infomation just ask.
Many Thanks
Mark