vicidial.org

Hello,

Is there a channel limitation within a vicidial cluster?

I have set up 19 survey campaigns with 50 channels each and it seems no matter what i do it will not go over 900 calls.

I have even tried 1 campaign with 1500 channels and it doesnt seem to get close to the number of channels i have selected.

Below is the setup that i am testing - there have been no issues with load avg or anything.

Database server: Supermicro (4) Opetron 6168 1.9GHZ Processors - 64GB Ram - LSI-9265-8I Raid - (6) Samsung 840 Pro SSD Drives in raid 10
Telephony Servers: (8) HP DL360 G5 2.3GHZ Servers - 8gb ram -
Telephony servers are set to 250 channels each @ 10 calls per second and are load balanced
Timing: Sangoma UT50 USB connected to each telephony server
G711U - Full AMD - No call recording
Version: 2.6-395a - Build 130221-1736
Asterisk Version 1.4.44-vici - Vicidial Redux v4.03

Any help is appreciated

increase calls per second (200 is good, but be sure your provider can handle it ... watch your Congestion/Cancels when you make the increase to ensure completed calls).

spread remote agents among servers.

We have a client with a large single Vicidial cluster that regularly places 2000+ concurrent calls, most of them on broadcast campaigns. So it is certainly possible with the right hardware, carrier and settings.

Any mysql setting recommendations for placing 2000+ concurrent calls?

install a vicibox 4.0.3 machine with "database only" during the configuration wizard. then have a look at the my.cnf.

and be sure you have 15KSAS or SSD drives and powerful DB server.

Check on all that stuff.
2x Intel(R) Xeon(R) CPU E5-2640 0 @ 2.50GHz
32GB of RAM
PERC H710P Integrated RAID Controller, 1GB NV Cache
4x 300GB 15K RPM SAS 6Gbps

I would recommend not using the PERC card, they never seem to be able to handle the IO fast enough. Get a real LSI MegaRAID Caching RAID controller instead. We've had multiple clients where changing just the RAID card to a MegaRAID gave them the ability to double in size. High quality SSD drives also help in improving performance. And if you can add more RAM that is always a good thing too.

I talked with the engineer and after giving him some numbers he said our RAID card was fine. I will continue to watch the numbers though.

Having experienced this in the past (and observed while it happened also), I would recommend getting the LSI card in-house for that "sudden need" moment. And keep an eye on your I/O wait time/bottlenecks.

If you're gonna RAID, at least be prepared to do it right (ie: have it handy to fix it if it suddenly needs it ...). Remember that these problems generally surface under Heavy Load. Which is the one time that any delay can be Very Costly.

Can you just swap RAID cards without rebuilding the server?

In theory the answer is yes if it's newer RAID cards, but it has never worked out that well in practice.

The safe solution is to back-up the server with an imaging program like CloneZilla and then give it a shot. If it bombs, re-image the server and away you go like nothing happened.

What we usually do is take a mysqldump from the server, move the file to another server, and then reload with vicibox on top of the new raid card. We then just re-import the database.