Network Usage - Packets being lost

[adam.greenhalgh1]

Hi all,

Device Received Sent Err/Drop
lo 424.17 MB 424.17 MB 0/0
eth0 1.17 GB 3.99 GB 0/0

Please see above and if you could direct me on the best way to move forward to prevent that amount of packets being lost.

We have a Snom320 connected to ViciDial via a conference. Every so often at random times - another call enters the conference without prior warning. Whilst the agent is still INCALL - The call that enters the conference is a voicemail message - sort of like the dialer has dialled another number whilst the agent is on the call but that call is not answered.

Could the above report present the issue we are facing with the Snom phone?

PS - the phones that are experiencing this issue are XP machines.

Much appreciated

[omarrodriguezt]

http://wiki.vicidial.org/index.php/VICI ... o_an_agent

[adam.greenhalgh1]

Thanks a lot Omar.

We are just waiting for the dialer to be attached to the NTP server that the main Server is on.

Hopefully this solves the issue


Ive done a Server report - please find the results below.

---------- TOTALS, PEAKS and AVERAGES
Total Calls in/out on this server: 228082
Total Off-Hook time on this server (min): 153306.13
Average/Peak channels in use for server: 5.6321 / 89
Average/Peak load for server: 15.3835 / 594
Average USER process cpu percentage: 2.1790 %
Average SYSTEM process cpu percentage: 0.7105 %
Average IDLE process cpu percentage: 96.6325 %

Nothing has peaked over 100% but just to be on the safe side

[omarrodriguezt]

You are welcome.
If the trouble shooting helped you, please fill this little poll http://vicidial.org/VICIDIALforum/viewt ... =2&t=27954

[adam.greenhalgh1]

No as of yet - but would you go over the report ive produced on the previous post. As im unsure if that is OK. As there has been some issues with voice or audio going to and from the prospect.
Cheers pal!

[omarrodriguezt]

??????????/

[adam.greenhalgh1]

Ill reproduce the report if needs be omar but replying '???????????/'. Is not helping anyone out mate!

---------- TOTALS, PEAKS and AVERAGES
Total Calls in/out on this server: 228082
Total Off-Hook time on this server (min): 153306.13
Average/Peak channels in use for server: 5.6321 / 89
Average/Peak load for server: 15.3835 / 594
Average USER process cpu percentage: 2.1790 %
Average SYSTEM process cpu percentage: 0.7105 %
Average IDLE process cpu percentage: 96.6325 %


Is this OK? Would i need to look into anything?

[DomeDan]

I totally agree.

Regarding that Server Performance Report, if this value "Average/Peak load for server: 15.3835 / 594" is correct then there is a problem,
but I got a small memory that it ain't normal linux load it is talking about, I will activate the "system performance" option here and take a look... (no its not linux load, its something else)

Login to the server when you experience the problem and run htop or uptime to view that load value

if you got a 4 core system then the load average should not exceed 4.0, first value is 1 minute average, next is 5 minutes average and the last is 15 minutes average.

So this problem only happens to XP machines? how many runs XP vs anything else?

[adam.greenhalgh1]

Dan ! Much appreciated for the reply buddy!

Im running 2 core system.

Around 50 agents 10 of which XP the rest are Win 7.

I wouldnt know how to run the htop or uptime? Or are they buttons? lol

Apologies for the novice response but - As you can probably tell I'm fairly new to the whole ViciDial scene

Adam.

[DomeDan]

If you got 2 of those Core(TM)2 Duo CPU E760 processors, then you got a total of 4 cores in your system. (because your signature says: "Processors 2") I'm I right?

htop and uptime is linux commands, log in to the system with ssh (use putty or cygwin on windows) and run the commands
or go to the physical screen connected to the server and log in there and run the command.

might just be a coincident that it happens to only XP machines so far...

if you are running 50 agents, database/web/asterisk on that processor then it got to be caused by a overloaded system
I think the recommended is to have 25 agents per asterisk server, so you should really move to a multi server environment.

but tell us a bit more about the system, the server and usage like how many logged in users, codec and stuff
and how many calls per day (the report you gave us shows a number by not the time span of that number)

and get the load average from the uptime command and print it here, both when you are experience the issue and when there is no problem

[adam.greenhalgh1]

Just a quick reply - ive not got ALL the information you asked for but. The time span of the report i produced earlier on was since day Z (Since we started using the dialler)

Ill get on the other info you asked for and get back to you.

Again Dan - Much appreciated!

[adam.greenhalgh1]

As this goes way beyond my scope of knowledge very quickly i hope your baring with me

I logged onto our server via Linux Shell - Which opened up a CLI for our server.

I typed uptime and got
14:51:44 up 19 days, 1:55, 1 user, load average: 0.03, 0.13, 0.15

Im assuming im doing something wrong as there has been ALOT more than 1 user!

I tried running the htop command and got Command not found.

Yes we have 2 virtual and 2 physical cores. I can post our entire hardware for our system - if thats any use?

"As for extra information you asked for.but tell us a bit more about the system, the server and usage like how many logged in users, codec and stuff
and how many calls per day (the report you gave us shows a number by not the time span of that number)"

Is there a report that will drag this information out? Or am i able to do it via the CLI?

I hope your still here Dan

Adam.

[DomeDan]

Yeah no problem, I bare with you

I typed uptime and got
14:51:44 up 19 days, 1:55, 1 user, load average: 0.03, 0.13, 0.15

Im assuming im doing something wrong as there has been ALOT more than 1 user!

"1 user" here means one used logged in to the linux operating system, and that is you who just logged in, so dont worry thats nothing wrong there.

"load average: 0.03, 0.13, 0.15" that's good, as I said before its:
1 min average: 0.03
5 min average: 0.13
15 min average: 0.15

and you said you had 2 virtual and 2 physical core, I guess you mean you got Hyper Threading enabled
we just count the physical ones. so that means that as long as those values does not exceed 2.0 then you are OK
but if the 15 min average is at about 1.75 then you should start to worry
if 1 min average spikes above 2.0 its usually no problem
but above 2.0 in 5 minutes average is quite bad

above 2.0 in your case means that processes will have to wait before they can get processed and that usually results in strange things happening in the vicidial system, like multiple calls in your case

information such as how many logged in users and codec and stuff needs to be drag from different places.

logged in users: the real-time report or just look in the room how many agents are sitting there and making calls.
dial method: is this an inbound or outbound callcenter, or some other kind, or a mix? ask the one who uses the system
codec: under admin page -> carriers -> should be in "account entry"
and the codecs for the phones, that can be on a few different places if its set in the admin webui, else you need to get to the asterisk cli, but never mind that for now.

number of calls per day etc can be pulled from reports,
explore the reports and you will find them and much more interesting stuff

but never mind that info can be useful later on, right now I would just like to see the load average value on the system when you are experience the issue you are having

[adam.greenhalgh1]

Mate you are a freaking godsend! haha - Ill have to get back to with the load average. Ill try and pull it at the peak of the day which in our case is around..11-12 o'clock.

So wont be able to pull until tomorrow.

You would want all the other information in one post - so that it can be compared against the time span. ?

Also - HOW DID YOU LEARN SO MUCH? Ive learnt alot so far but its only using it as such. Not really installing/configuring. I configure campaigns - phones - agents - lists- leads - troubleshooting. I could probably benefit from having a tutorial or a form of course to assist in my learning curve. But thats by the by.

[adam.greenhalgh1]

I can shed some light on the whole multiple call entering the conference.

The dialer has rung a number connected the number with the agent.

The dialer then rings the same number but in the back ground - and then displays the VM message. (because the number is already in use)

Im assuming its all down to the time servers that all our systems are running from. As they arent officialy running from the same NTP server - that WILL prevent problems.

We connected our main server and PBX to an NTP but as im a novice i cannot hook up the ViciDial server to the same NTP that our main server is on.

If you could shed some light on how to do that? Or just reply with...To advanced...if you think it would be out of my scope. Im assuming it will be - but hey ho..Heres HOPING!

Cheers..

Adam.

[DomeDan]

Haha, godsent ^^ nah

Yeah grab the load average at the peak, but also right when someone complains that they had this issue where there are too many calls in one conference, because it might not be related to the load, that's what we have to find out.

the other information is not really related to a time span, its just some basic information of the system and usage thats good to mention when troubleshooting.

I've learnt from the old-timers here, when I started using vicidial I read countless topics from this forum,
I had good knowledge with linux before starting to use vicidial and that helped a whole lot. linux knowledge is always good have! its very useful.
And of course, vicidial is free and open source so anyone can look at the code and learn from that, and do debugging with the code when there is a problem with something.
so learn by exploring and hacking is how I do it!
the vicidial team got some training program I've seen, never tried it but its worth mentioning. but I guess you are located in Australia and you might need to travel to Florida, don't know, haven't checked it out,
ahh, Australia... had plans to travel there, going for gold detecting

[DomeDan]

Oh, it is a multi server setup?

please tell us about the system, what servers are there and how are they connected?

The dialer then rings the same number but in the back ground - and then displays the VM message. (because the number is already in use)

That could also be related to the dial plan, post the dial-plan under from "admin -> carrier"

also posting the output from the asterisk cli when this happen would be very usefull

[DomeDan]

Setting a different time server is quite easy,
I haven't tried Centos (which Goautodial uses I think) but I guess its the regular way to do it.

You log in to the system and edit the file /etc/ntp.conf in your case the application nano might be the easiest texteditor:
nano /etc/ntp.conf
there will be some row containing "server ntp.server.com" and there is probably a few of those, just edit them to point to the server you want to use
you can specify several, one per line:
server 0.ntp.server.com
server 1.ntp.server.com
server 2.ntp.server.com

and restart the service like this:
/etc/init.d/ntp restart
if there is errors then its ok, or if centos does not like that then try it using the service command:
service ntp restart

[adam.greenhalgh1]

Dan - Im from the good ol' England!

Still live here aswel And the need for the training isnt a high priority where i work - as there is other forms of calling. Besides the very practical and easy to manage systme that is ViciDial. Like manual dialing directly from hard phones. All VOIP.

I will post my hardware spec on here tomorrow buddy. Because as for now im off home - After a long hard day on the grind! I would provide the information from home but i cant log in remotely as like i said its not a high priority the dialer.

However im very interested in ViciDial And Diallers in general, but in terms of Linux i have a very limited..Infact we will call it NO knowledge! I will be borrowing books reading on forums and takingonline tests and i will learn linux in different forms. As i believe it changes very regularly lol! And most dialers and pbx's are all run off Linux.

It has been nice taslking to you Dan - We will talk more tomorrow. Cya pal

[DomeDan]

Oh, its not only Australians that use words like 'mate' and stuff

there are many variations with linux, but the main gnu/linux system is the same (works the same)
so if you pick up a book from around year 2000 it will still be very much useful if it keeps to command line and topics like text-editing, input/output, pipes, and stuff like that,
the rest is learning by doing, installing packages using the system package-manager, dont need a book for that, and you got the linux man-pages!

Nice talking to you too, cya!

[adam.greenhalgh1]

Hi dan, hope your well buddy.

Yeah i think its more on a personal level - "MATE"
We are if it was professional i would no way say "mate" haha.

Sounds good mate - My linux knowledge started yesterday! Cheers dan

Are you from Sweden then yeah?

Also would you recommend any books for me to read? - Baring in mind i want an understanding a basic understanding for now. The term walk before you can run springs to mind.

Cheers Dan!

[DomeDan]

I'm all well thank you, and yeah I'm from Sweden,
I think the book I read was in Swedish, but did some research and found one that might be good: http://linuxcommand.org/tlcl.php seams to have all the stuff, I want, a book about gnu/linux should have.

Did you see my message regarding changing ntp server?
and tell me a bit more about the system, I thought it was a one-server-setup, maybe its some other reason you have this issue..
but the load should still be checked the next time you experience the issue

[adam.greenhalgh1]

Hi Dan,
Thats good to hear buddy.

Yeah ill have a look at that link today buddy - Cheers.

And no issues as of yet Dan, The changes to the NTP server are being addressed by our external support agents. But thought with guidance i may be able to do it myself.

It is just one server, that has GoAutodial installed on them. Well one main and a back up.
We have an Asterisk - Apache - My SQL - But its held on one server. With another to back it up.

But as of yet the issue has not arrised again. And i think we will be eradicating the XP machines within the next few weeks. So HOPEFULLY all things go well and we dont have this issue. However the server will get connected to an NTP server just for clarification.

[adam.greenhalgh1]

This has now been fixed. - The problem was a faulty mouse on the agents machine.

ROOKIE! :/