Lead information loading delay

[konextu]

Not sure if anyone has seen this issue as we are starting to investigate it now but we are having the following issue, and it seems very random.

Load Averages:
MySQL Server: 0.18, 0.29, 0.37
ViciWeb Server: 0.03, 0.08, 0.03
ViciDial dialer Server: 0.75, 0.53, 0.50

Dialer has 24 gigs of ram with 18 gigs free
MySQL Has 16 gigs of ram with 12 gigs free
Web Server Has 4 gigs with 3.5 gigs free.
24 agents logged into the system
Outbound SIP calling Through BinFone
Inbound IAX calling through BinFone

Here is the problem:

On random outbound calls the agent will get the call and after about 15 to 30 seconds the information will appear on the vicidial agent screen and she will get the popup and the screen will show live call, this has been known to go on for up to 45 minutes at a time and then just stop happeneing. It doesnt affect all agents just random ones at random times. I've check the log files on the web server to no avail.

Not sure if anyone else has ran into this sort of problem?

Thanks

[okli]

Maybe check with mtop if there is a slow query when this happens? This may give you clues for the root cause.

Try also the search option in this forum, I've seen topic(s) in the past about the same issue.

[Geil21]

I had this problem too

I found that MySQL was hanging when the hopper load script was running and as a result the data was very slow to get to the agent ( and it was slow to grab the next lead to dial too )

I was using filters at the time and added some indexes to the vicidial_list table on the fields that I was filtering on to get the queries to run faster

After adding the indexes there was no more lag

Things I would check are:
is your hopper level too high?
are you using any custom filters on non-indexed fields?
if you watch top on your db server what is the cpu usage for mysql when it's happening?

[konextu]

It could very well be an hanging hopper script. I did some quick searches but did not find any results, in the event that there ARE other posts directly related i will apologize. Thanks for the heads up on sql being a lagging b*tch, i'm going to dig some more.

[mflorell]

I would suggest running, and watching, "mtop" on your database server and seeing if any patterns emerge when this stuff happens.

[konextu]

one other point to add, we have over 1.5 million leads in vicidial_list. Think this could be the problem?

[Geil21]

most definetly

our problems started happening when vicidial_list got to about 1 million

a good index will beat all the RAM in the world
just look at the fields included in the WHERE statement in the hopper loading script then create individual and combined indexes on those fields

of course this increases overhead but i'm sure you've got the space based on your stats

[mflorell]

We have clients that have had over 12,000,000 leads in their vicidial_list table without any issues, but then again they had a dedicated database server. The limit really depends on how your system is set up and the hardware specs on your database server.