PROB : Many connected calls to one agent

[dev_4901]

Hi,
Finally I have got the x206 server up and running with VICI Dial and all other requirements.

There is one issue which is causing major headaches here and I require advice to get over this.

There are 8 agents which log in into only one campaign at a time. Once the agents are connected to a live call, new calls are still coming to the already connected agent. Due to this, during a course of 30min or so one agent is connected to 3-4 customers at one time. We got no option but to hangup all the calls one by one by the HANGUP link in the SHOW CONFERENCE DETAILS.

Asterisk - 1.2.10
ASTGUI Client & VICI Dial Package - 1.1.12

Plz advice.

Dev SInghal

P.S : I noticed one thing, dont know whether its relevant or not, in the Asterisk CLI screen the Manager '*CRON' is now logging ON & OFF from 127.0.0.1 (loopback adaptor) but while I was testing on a different machine it used to be the IP of the machine.(192.168.x.x).

[Op3r]

Can you tell me your autodial level?
Whats your server specs?
Codec?


---------------------------------------
Suggestion:

try lowering your dial level.

ill know more and more people can help you if you answer those questions above.


Best regards,

Edwin

[mflorell]

What is the loadavg of the server?

You should be using the real IP address for the server not the loopback.

Are the web server and mysql on the same machine as Asterisk/VICIDIAL?

[dev_4901]

Thanx guys for the replies.....

Server is an IBM X-206 series server with Intel 3.0 Ghz , 1 GB RAM , 70 GB SATA.
Codec used - g723

Auto Dial Level is 1.2
Hopper Level is 50

Yes, the web server and Mysql are on the same server. I dont think 8 agents should cause so much load so as to separate them.

The IP addresses are setup correct. There are two NIC's one on local IP and the other NIC has real IP.
How do i get the manager (*CRON) to check on the local IP 192.168x.x instead of loopback adaptor?

Need help fast..........

Thanx in advance.

Dev Singhal.

[mflorell]

What is the loadavg of the server?

[dev_4901]

The Server Performance report is not working............

Also, how do i get the manager (*CRON) to check on the local IP 192.168x.x instead of loopback adaptor?

Plz advice.

Dev Singhal.

[dev_4901]

I found a list of commands on this forum, and I ran those. I'm putting up the output of those, if they can help(I HOPE)

root@localhost:~# /usr/bin/cat /proc/stat
cpu 82739 0 18333 346036 13194 602 4292 0
cpu0 50295 0 11103 159768 6981 602 3857 0
cpu1 32443 0 7230 186267 6213 0 435 0
intr 7151198 2326406 10 0 0 0 0 3 0 2325093 0 0 0 607 0 14 1152506 785898 560661 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
ctxt 7044917
btime 1156792865
processes 39789
procs_running 3
procs_blocked 0

root@localhost:~# /usr/bin/cat /proc/loadavg
0.87 1.05 0.79 3/149 8108

root@localhost:~# /usr/bin/free -m -t
total used free shared buffers cached
Mem: 1010 467 543 0 12 331
-/+ buffers/cache: 123 887
Swap: 1960 0 1960
Total: 2971 467 2503

root@localhost:~# /bin/ps -A --no-heading
1 ? 00:00:00 init
2 ? 00:00:00 migration/0
3 ? 00:00:00 ksoftirqd/0
4 ? 00:00:00 migration/1
5 ? 00:00:00 ksoftirqd/1
6 ? 00:00:00 events/0
7 ? 00:00:00 events/1
8 ? 00:00:00 khelper
9 ? 00:00:00 kthread
12 ? 00:00:00 kblockd/0
13 ? 00:00:00 kblockd/1
14 ? 00:00:00 kacpid
157 ? 00:00:00 pdflush
158 ? 00:00:00 pdflush
160 ? 00:00:00 aio/0
159 ? 00:00:00 kswapd0
161 ? 00:00:00 aio/1
748 ? 00:00:00 kseriod
819 ? 00:00:00 kirqd
824 ? 00:00:02 kjournald
882 ? 00:00:00 udevd
1274 ? 00:00:00 syslogd
1278 ? 00:00:00 klogd
1638 ? 00:00:00 shpchpd_event
1669 ? 00:00:00 khubd
4309 ? 00:00:00 inetd
4347 ? 00:00:00 sshd
4359 ? 00:00:00 crond
4361 ? 00:00:00 atd
4364 ? 00:00:00 sendmail
4367 ? 00:00:00 sendmail
4370 ? 00:00:00 acpid
4378 ? 00:00:00 httpd
4379 ? 00:00:06 httpd
4380 ? 00:00:07 httpd
4381 ? 00:00:07 httpd
4382 ? 00:00:07 httpd
4383 ? 00:00:06 httpd
4385 ? 00:00:00 gpm
4387 ? 00:00:00 mysqld_safe
4391 ? 00:00:00 in.tftpd
4419 ? 00:00:52 mysqld
4425 ? 00:00:00 ntpd
4489 ? 00:00:02 screen
4490 pts/0 00:07:53 asterisk
4496 pts/0 00:00:00 mpg123
4497 pts/0 00:00:00 mpg123
4510 tty6 00:00:00 agetty
4511 ? 00:00:00 kdm
4527 ? 00:00:01 X
4528 ? 00:00:00 kdm
4560 ? 00:00:00 kdm_greet
4584 ? 00:00:00 screen
4585 pts/1 00:00:04 AST_manager_lis
4590 ? 00:00:00 screen
4591 pts/2 00:00:04 AST_manager_sen
4595 ? 00:00:00 screen
4596 pts/3 00:00:05 AST_VDauto_dial
4602 ? 00:00:00 screen
4603 pts/4 00:00:01 AST_VDremote_ag
4604 ? 00:00:00 screen
4605 pts/5 00:00:41 AST_update.pl
4622 ? 00:00:06 httpd
4655 ? 00:00:06 httpd
5116 ? 00:00:08 httpd
5176 ? 00:00:06 httpd
5299 ? 00:00:06 httpd
5354 ? 00:00:07 httpd
5634 ? 00:00:05 httpd
5759 ? 00:00:05 httpd
5843 ? 00:00:06 httpd
5954 ? 00:00:06 httpd
12544 ? 00:00:05 httpd
22305 ? 00:00:02 httpd
22871 ? 00:00:02 httpd
27511 ? 00:00:00 sshd
27614 pts/6 00:00:00 bash
8858 ? 00:00:00 AST_CRON_mix_re
10139 ? 00:00:00 AST_manager_kil
10452 pts/2 00:00:00 AST_send_action
10487 pts/0 00:00:00 VD_amd.agi
10490 pts/6 00:00:00 ps

[mflorell]

That is not a bad loadavg, is that when you are experiencing the problems?

To activate the server stats, change the $SYSPERF = 1; in AST_update.pl at the top of the script.

To change the login IP just edit the AST_SERVER_conf.pl file in /home/cron

[dev_4901]

Yes, the stats I have posted are when all the agents are logged in and the problem is there.

I have $server_ip = '192.168.x.x' Local IP of the server
$DB_server = 'localhost'

PLz advice and thanx in advance.

Dev.

[mflorell]

You will need to look through your /home/cron/action_full.2006-08-XX file and find the failed Hangup actions in there and post a few of them.

[dev_4901]

|
2006-08-25 5:44:20|0|188
|
Action: Hangup
Channel:

|
2006-08-25 5:44:20|1|188
|Response: Error
Message: No channel specified

|
2006-08-25 5:44:20|2|188
|Response: Goodbye
Message: Thanks for all the fish.

Is this like something what I should post?
There are many like these

[mflorell]

Yes, that's what I wanted. Can you see if your AST_update.pl has stopped running at any time for some reason?

Can you post the output of "screen -r"?

[dev_4901]

root@localhost:# screen -r
There are several suitable screens on:
6767.ASTlisten (Detached)
6773.ASTlisten (Detached)
6779.ASTlisten (Detached)
6788.ASTsend (Detached)
6803.ASTVDauto (Detached)
6811.ASTVDremote (Detached)
6815.ASTupdate (Detached)
Type "screen [-d] -r [pid.]tty.host" to resume one of them.

[mflorell]

You seem to have 3 ASTlisten screens running somehow. Can you kill 2 of them and see if that helps things for you?

[dev_4901]

This is the output of ps -x

Which ones do i kill?

Would I have to do this every time? (I HOPE NOT)

6767 ? Ss 0:07 /usr/bin/SCREEN -d -m -S ASTlisten /home/cron/AST_manager_listen.pl
6768 pts/0 Ss+ 0:49 /usr/bin/perl /home/cron/AST_manager_listen.pl
6773 ? SNs 0:05 /usr/bin/SCREEN -d -m -S ASTlisten /home/cron/AST_manager_listen.pl
6774 pts/1 SNs+ 0:57 /usr/bin/perl /home/cron/AST_manager_listen.pl
6779 ? SNs 0:05 /usr/bin/SCREEN -d -m -S ASTlisten /home/cron/AST_manager_listen.pl
6780 pts/2 SNs+ 0:56 /usr/bin/perl /home/cron/AST_manager_listen.pl
6788 ? Ss 0:01 /usr/bin/SCREEN -d -m -S ASTsend /home/cron/AST_manager_send.pl
6789 pts/3 Ss+ 1:47 /usr/bin/perl /home/cron/AST_manager_send.pl
6803 ? Ss 0:20 /usr/bin/SCREEN -d -m -S ASTVDauto /home/cron/AST_VDauto_dial.pl
6804 pts/4 Ss+ 3:48 /usr/bin/perl /home/cron/AST_VDauto_dial.pl
6811 ? Ss 0:08 /usr/bin/SCREEN -d -m -S ASTVDremote /home/cron/AST_VDremote_agents.pl
6812 pts/5 Ss+ 1:15 /usr/bin/perl /home/cron/AST_VDremote_agents.pl
6815 ? Ss 0:00 /usr/bin/SCREEN -d -m -S ASTupdate /home/cron/AST_update.pl
6816 pts/7 Ss+ 16:17 /usr/bin/perl /home/cron/AST_update.pl

Thanx in advance.

Dev.

[mflorell]

just kill any two of the AST_manager_listen.pl processes.

[alexjones]

hi,

I have the same problem. Can you help?

[mflorell]

Please post your loadavg when the problem is occuring.

Also, describe your setup in more detail including how many agents, what kind of trunks, whether you have Asterisk/Apache/MySQL all on the same server, etc.

[dev_4901]

Matt,
I've got the hang of it now. Sometimes when I start the server, more than one instance of AST_listen starts. There is only thing that I can't figure out, why this occurs?
Whenever this happens I kill all except one and then the calls go smoothly.

Thanx again. Shall keep you posted in case I require further help.

Cheers,
Dev Singhal.

[mflorell]

On some servers the listen script would launch a second or third instance if Asterisk was restarted at certain times. This has been fixed in the 2.0.1b1 release.

[alexjones]

Please post your loadavg when the problem is occuring.

Also, describe your setup in more detail including how many agents, what kind of trunks, whether you have Asterisk/Apache/MySQL all on the same server, etc.

I dont know the load average, the script is not working.
We have a dell SC1420 xeon HT 3,2 Ghz, 1 GB RAM, 2 x 160 GB hd RAID.
Now we have 20 operators, we have SIP trunks with softphone xlite.

All the software is installed on the same server.
Tell me if you need more from me.
Thank you.

[mflorell]

cat /proc/loadavg

[alexjones]

cat /proc/loadavg

Forgot to mention that we use Gentoo not Slackwere, but i dont think thats a problem.

I will post the load average in a short time when i will have my agents on the dialer.

[alexjones]

cat /proc/loadavg

This is what is showing me when i put the command:

4.36 4.05 2.89 1/270 13995

[mflorell]

Unless you are running a Dual Xeon with Hyperthreading enabled or a quad processor system your load is way too high.

Please post your system specs.

Are you recording all calls?

[alexjones]

Unless you are running a Dual Xeon with Hyperthreading enabled or a quad processor system your load is way too high.

Please post your system specs.

Are you recording all calls?

We have a Single 3.2 Ghz HT Xeon with 1 GB RAM and 2 hd in raid of 160 GB.
We dont use any digium hardware because we use only sip, would it help to have something for the timing or i dont know for something else?

We are not recording calls at this moment.
What are your suggestions?

[mflorell]

You are running at over 200% load which is probably causing your problems. If you are using a codec like G729 you should try having a gateway machine do transcoding to G711 which will greatly reduce load. Other than that VICIDIAL works great load balanced across multiple servers.

[alexjones]

You are running at over 200% load which is probably causing your problems. If you are using a codec like G729 you should try having a gateway machine do transcoding to G711 which will greatly reduce load. Other than that VICIDIAL works great load balanced across multiple servers.

Yes we are using G729. From xlite(softphone for every agent) to vicidial server i think we use ulaw and from server to telephony provider we use G729.How do i setup that machine to do the transcoding?

[mflorell]

You could buy an appliance like a phonecodes gateway or you could use an Asterisk server for it.

For the Asterisk transcoder you just need it to have Asterisk with the G729 codec libraries, no timer needed. You can have that server and the VICIDIAL server register to each other over SIP or IAX (disallow=all allow=ulaw) and then pass all calls out to the gateway server and on the gateway server turn off logging and most of the Asterisk modules and just have it pass all calls to your SIP provider.

[alexjones]

You could buy an appliance like a phonecodes gateway or you could use an Asterisk server for it.

For the Asterisk transcoder you just need it to have Asterisk with the G729 codec libraries, no timer needed. You can have that server and the VICIDIAL server register to each other over SIP or IAX (disallow=all allow=ulaw) and then pass all calls out to the gateway server and on the gateway server turn off logging and most of the Asterisk modules and just have it pass all calls to your SIP provider.

And what configuration do you recomend for both of the machines? and what digium card(s)?

Can t i do it only with one machines?

And if i do this i will not have any problems after ?

What do i have to do to have 30 agents with dial level at least 2?

[mflorell]

When you get to larger setups you need to start separating tasks out to multiple servers.

For 20 agents if you moved the database and web server off to another machine you might be able to do all 20 agents on your VICIDIAL server.

For 30 agents you might want 2 VICIDIAL servers and one DB/Web server

For 50 agents you might want 3 VICIDIAL servers 1 DB server and 1 web server

Upto where we are with 120 agents at one location with 7 VICIDIAL servers, 1 DB server and 2 Web servers.

But the capacity very much depends on what you are doing with the system.

As for Zaptel cards, on the VICIDIAL servers you will want to have an X100P for timer. If you are doing a transcoding server no zaptel card is needed.

[alexjones]

can you give me a code update to fix this?

[mflorell]

What are you looking to have fixed exactly?

[alexjones]

What are you looking to have fixed exactly?

The problem with the calls mixing toghether. You said you fix this in the new version. Like an agent while is in call he receives another call even if he is already speaking with someone. Or sometimes agents can hear themselvs in the same conference.

[mflorell]

2.0.1b1 has a hardcoded option in vicidial.php to not re-issue a sessionID (which you would have to reset at the end of the shift with the AST_reset_mysql_vars.pl script)

As for more than one call in a session, that is almost always load-related and there is no coding around that.

[alexjones]

2.0.1b1 has a hardcoded option in vicidial.php to not re-issue a sessionID (which you would have to reset at the end of the shift with the AST_reset_mysql_vars.pl script)

As for more than one call in a session, that is almost always load-related and there is no coding around that.

So the only option is getting a more powerful server and reduce the load, i cannot fix this in another way?

[mflorell]

It would probably be cheaper to split functions off to other servers actually. There is no way to program around server load on Asterisk/Mysql/Apache so adding more hardware is your only option.

[Op3r]

Kindly check the load balance txt that was included when you dowloaded the astguiclient zip file.. anyway here's a tip

We got 30 agents on our call center. All calls are recorded.

Here's my setup.

2 Asterisk server - Load balanced (one with x100p and one with just ztdummy)

1 Mysql/webserver

Never experienced cross lines and such.

[artimus]

We are having this issue as well. One agent will get two calls at the same time.

I notice the load spike to about 4seconds and then back down under 1 second. I'm not sure what is causing it to spike. An initial look seems to be either asterisk or httpd. We have already moved mysql to another box.

Why exactly is this happening? I have a guess as to what is going on but I have not really checked yet.

I am guessing that vici does a sql statement to see which agent is available. Then it passes the call to that agents meetme. Finally, vici updates mysql to say that the meetme is now in use.
If this is the case, then vici should create a random id number and try to insert that into the database. It should then select from the database limit 1, and if the random id matches then it could connect to the meetme. Otherwise, it was too slow, and should try again to find another available meetme. Am I on the right track here?

[mflorell]

Actually VICIDIAL does already do an UPDATE ... LIMIT 1 to lock in an agent, not a select so that is not your issue.

If you have a web server or DB on any of your Asterisk/VICIDIAL servers I would suggest moving them off to a dedicated server.

What is your loadavg when it spikes?

What kind of CPU do you have on your server?