PROB : Many connected calls to one agent

[artimus]

What do you suppose the problem is. I understand that the problem is caused by a high load, but this type of problem should never happen no matter what the load is. I would rather the server shut asterisk down before allowing it to connect multiple people to a conf. Do you know where in the code the problem exists, or why the problem exists?

Current load right now is 0.95. The spike yesterday was 14. I normally get nervous when I see it get close the 3.


This is a dell pe1750. 2.4Ghz Xeon, w/ 2G ram. There were about 10 agents logged in during the spike. The heaviest processes where apache, asterisk, and AST_update.pl.

We started building a new server this morning so that we can push apache off as soon as possible. MySQL was moved off a while ago to box with more hd space.

[mflorell]

Aside from my dislike of Dell servers, a load of 14.00 on a single CPU system will create all sorts of problems for VICIDIAL, and just about everything else.

Have you run the server performance logging and report?

I would recommend moving some process off of that server.

[Op3r]

Load balance it man. It always comes down about the load. I'd rather be wasting resources than hogging it when it comes to VICIDIAL. Cramming seats into a server is very very foolish.

But for 10 seats that should be ok if youre not recording your calls. Have you happen to install the php-eaccelerator?

and please get a respectable server from supermicro. Hating dell servers is a sport of its own. If dell wants to disprove that their server sucks they are more than welcome to send some servers for testing

As for crosslines and such. Always double check how your agents logout.