Realtime report agent in queue for longtime

[vijaymuddu]

Hi

I am facing issue where in, Real-time report shows agent in queue but actually agent is in call and also multiple calls hits the agent whose status is in queue

please anybody facing same issue or has solution for the problem


VERSION: 2.4-364a
BUILD: 120409-1136

[MJCoate]

Hello,

This sounds like network and/or agent browser problems:

1) If you have packet loss or bad latency, you will see this sort of thing. The packets that would normally be updating the server with the agent status get lost and you end up with inconsistent data on the Realtime Report.

2) You can also experience the multiple call problem if the agents web-browser is messing up and retransmitting packets when it shouldn't. The dialer gets multiple signals that the agent is ready when it should only get 1 and so it sends them multiple calls.

You can look at the Agent LAGGED Report in Reports > Admin Utilities to see if the network is a problem. The more agents that show up on it the worse your networking issues will be. To see if there are retransmits, go into the user stats for one that is having the problem. If you see multiple LOGIN events without corresponding LOGOUTs preceding them, then that agent's browser is misfiring. You can try rebooting the machine, disabling extensions and avoid going to sites that use AJAX to mitigate that.

[williamconley]

Um ... sounds to me much more like an overloaded server or crashed table.

Also, you listed goautodial 3.3, but vicidial 2.4. one of those is likely very wrong.

You should consider upgrading to the latest version of Vicidial if you are actually in 2.4 (we're presently in 2.12).

Also consider re-installing with Vicibox 7.X.X (whatever's out when you rebuild). Lots of new toys, more stable environment. Upgradeable and clusterable.

But check on those broken tables first. Could very well cause the issue.

[vijaymuddu]

Hi

My server top output shows mysql 129% cpu usage and asterisk using 31.9%, config is i5 processor, 8 gb ram

Code: Select all

 4531 mysql     24   0  433m  64m 4716 S 129.2  0.8 621:25.13 mysqld
 4798 root      15   0 1746m  58m 6180 S 31.9  0.8  84:26.84 asterisk
 5880 root      16   0 75364 9.9m 2880 S  4.3  0.1   2:31.74 AST_update.pl
10645 root      15   0  149m  89m 2772 S  2.7  1.1  10:16.64 AST_VDadapt.pl
 8360 apache    15   0  307m  19m  12m S  1.3  0.2   0:12.01 httpd

[edited by williamconley for legibility]

can anybody suggest how can i reduce mysql load, might be this causing agents to go on queue status

agents 40
call type : blended
predictive dialing with ratio set 1.5
IVR on inbound calls
vicidial version upgraded to : Version: 2.12b0.5
DB Schema Version: 1456

[williamconley]

Please remember to always list the FULL vicidial version With Build. Makes a difference for everyone (not just you) to link those with similar problems using the same build ...

Onward!

When your system overloads, it can cause mysql to overload. Mysql may or may not be the actual cause of the overload.

If your system has been in use for a long time, you should be archiving (and perhaps pruning) your log tables. There are processes in the /usr/share/astguiclient perl scripts folder for this purpose. They are also usually included in crontab -e but commented out for ease of activation. Read the comments in the scripts for archiving to see what they do. Some of the log tables have no value if you are not troubleshooting and others are actually used in Reports, so don't just go nuts and start deleting entries.

If any of your database tables have millions of records, it's entirely possible mysql is beginning to have troubles with containment. If any have Tens of Millions ... even more so. Look at how to reduce those table sizes.

Next up: there's a mysql configuration file aptly named my-big-vici.cnf (or similar). It has settings for high-volume systems which may come in handy.

You can check active queries during overload times to see if there are specific queries causing issues (such as time zone lookups or filters). "show processlist" is a very handy mysql command. You can also use phpMyAdmin or mytop or mtop (or workbench for mysql).

All sorts of things can overload Vicidial. But one of the biggies: 40 live agents autodialing on a single server. You may need to Cluster your system. Which makes it even more likely you'll want to use Vicibox for your next install! It clusters automatically. (Thanks Kumba!)

[vijaymuddu]

Hi

I am still facing same issue agents showing in queue for longtime and mutilple call hitting agents at a time

have tried to increase the server ram and also used huge my.cnf config for mysql, still the problem persists, mysql load and asterisk load is very low still there is issue, also i can ping my server very smoothly no delay

Code: Select all

4678 root      15   0 1352m  39m 6132 S 36.6  0.2  22:40.25 asterisk
4423 mysql     15   0 9548m 1.1g 4668 S 19.6  6.8  39:45.65 mysqld

please can anybody guide what can be the cause

[williamconley]

uptme and htop both show Average Server Load. This is an important value which should never exceed the number of cores you have in your server. In fact, it's best if it never exceeds 50% of the cores in your server for best performance. So if you have an eight core server, average server load should stay below 4.0 for all three values in uptime's average server load.

In many cases, the only way to reduce the average server load is to add another server to take some of the load. Before going this far it's often useful to reduce waste. Turn off unnecessary logging. Don't run reports or load leads during dialer operations. That sort of thing.

[vijaymuddu]

hi William

Thanks for feedback

I think problem get sorted, it was issue with google chrome we have switched all agents to Firefox which was in-build in ubuntu os, now there are no queue status on real-time report all no mutiple calls hitting agents

But still i am surprised why its happening with google chrome