vicidial.org

Hello Vicidial Forum,

My Machine:

Asterisk 11.25.3-vici built by abuild @ lamb68 on a x86_64 running Linux on
ViciBox v.8.0.1
VERSION: 2.14-675a
BUILD: 180520-1749

My Issue:

Defined dial level as 4:1 Dial method Ratio (Checked that a 100 times), when I have around the 20 agents I seem to have only 22 calls basically as if it was 1.1:1 which is weird.

What could be the possible causes for this and what would you try?

What I tried:

Reindexing the database - basically the big tables call_log , vicidial_log , vicidial_list - didn't do much it seems...


Please Advise.

Shoot just saw now that the server system load 2143 - 100%
how to do about this?

Dial Method?

EXviciman wrote:Hello Vicidial Forum,

My Machine:

Asterisk 11.25.3-vici built by abuild @ lamb68 on a x86_64 running Linux on
ViciBox v.8.0.1
VERSION: 2.14-675a
BUILD: 180520-1749

My Issue:

Defined dial level as 4:1 Dial method Ratio (Checked that a 100 times), when I have around the 20 agents I seem to have only 22 calls basically as if it was 1.1:1 which is weird.

What could be the possible causes for this and what would you try?

What I tried:

Reindexing the database - basically the big tables call_log , vicidial_log , vicidial_list - didn't do much it seems...


Please Advise.

Ratio as I mentioned

Wait a minute. "Dial level" meaning ... a value you see on the screen when modifying the campaign named "Auto Dial Level" or "dial level" as in "how many calls are being made per agent according to your calculations"?

If you mean the former, the only way the auto dial level changes if you change it or one of the Adapt settings is in use for Dial Method.

If you mean the latter, then it's not a single thing, it's a combination of things and you have to check them all one at a time to see what is happing. Items to check

* Dial Method
* Auto dial level
* Available only tally
* Leads available to be dialed
* Hopper level
* Leads in the hopper
* Agents in READY state according to Real Time screen
* Maximum Trunks
* Maximum Calls Per Second
* Carrier Stats (to see if calls are CONGESTED or UNANSWERED)

Dial Method = 4
Available agent = 10
Available on Tally currently N
Leads avaialble - not gonna state but there are enough for 20 agents for the next month
Leads in the hopper 100+
Agent in READY state according to real-time = 10
Maximum trunks = 100
Maximum Calls Per Second = 5
Carrier stats :

Code: Select all

CONGESTION = 7.3% (LAST 6 HOURS) ,
CHANUNAVAIL = 5.4% (LAST 6 HOURS)
NOANSWER 0.0% LAST 6 HOURS

[edited - formatting]

Dial method is NOT 4.

You didn't specify "former or latter".

Your "Carrier stats" was very limited. It would be better to post the whole thing. Nothing personal in those stats.

Unless you're saying you has NO answered calls in the last six hours?

Is your auto-calls table full of calls that are not in progress?

Code: Select all

asterisk -rx "core show channels"

does the "state" column with "ring" calls in the result this agree with "ringing" in Vicidial? (within 10% is normal)

Do you have an "s" extension in at least one of your Carrier's Dialplan Entry fields? (this catches impossible phone numbers and kills them instead of ignoring them and leaving them as "ringing", blocking a new dialout with a fake dead phone call!)

Hello,

We have some progress on our end, we did some defragmentation + found some long queries which were slowing the calls (I will PM you stats in a bit as well)

Now the new phenomena is burst of calls. First there are no calls the agents waits , and then after 1-2 minutes suddenly a burst of calls with a lot of drops , but after that wait time again for the agent and that goes on.....

On the "core show channels" I checked and for example if the number on real-time main report is 51 calls ringing , in the "core show channels" there are only 15 calls with status "ring". I checked on more occassions and had similar cases where the number of calls with a "ring" was more than half of what shows on the real-time-main-report (which refreshes every 4 seconds).

I can show post examples of "core show channels" as well if you think it can help

Don't PM me. I only do direct contact with Clients (ie: paying customers). The company will not allow me to do offline support for anyone else. Free stuff is visible (and usable) for all.

Try two things:
1) Change the Drop Call Seconds to "0" (zero). This has the unusual side effect of causing new calls to generate at the moment the old call terminates instead of "try to find an agent ... watch this process later".
2) Calls Per Second: Set this number to 1 CPS per 10 agents logged in and Ready (assuming your dial ratio is 4:1 or less). This will stretch out the dialing. Instead of all at once, you'll get one per second and by the time the last one is dialed, your first answer may have been resolved. So another answer will get an agent instead of dropping. Smoothes out the outbound dials to avoid "bunching".

However: 51 calls ringing but only 15 calls "ring" sounds like you may need to institute the "s" diaplan entry in a generic (fake) carrier to catch and kill bad numbers.

Code: Select all

exten=> s,1,AGI(agi://127.0.0.1:4577/call_log)
exten=> s,n,NoOp(No Dial Pattern Matches This Extension)
exten=> s,n,Hangup

Only put this in ONE carrier. I recommend a generic carrier so it won't get deactivated. Remember that all your carrier dialplan entries actually end up in "default" (all in the same place) so there can not be a duplicate. Only one of the dupes will actually be in the final dialplan.

Hey William,

Got it. Applied the above just now aka Set Drop Call Seconds to 0 , CPS on the server to 1, I set an active catchdups carrier within the carrier section in the gui for "s" dialplan (hope that what you meant)

Also added RAM + CPU.

Couple of questions I have:
1) Setting Drop Call Seconds to 0 - This means basically any calls unanswered is dropped immediately ?
2) You assumed we go 4:1 or less , what happens if we want to increase the auto dial level , how should we go about the CPS?
3) Not sure I quite understand how does this "s" dial plan generic carrier works , would be happy if you could explain.
4) Is there an advantage of adding multiple servers to make a vicidial cluster instead of one machine?

Also forgot to mention. Solved the problem.

The main issue turned out to be that we had an overflow of leads & lists on the campaign. (You can see that with long query log of mysql) basically each time the hopper is taking from the leads on the camp he went through all of the leads X the amount of agents which from a certain threshold of agents started to slow the database with a lot of big queries.

The solution was to reduce the amount of leads & lists on the campaign.