Upgrading ViciBox v.8.0 to Asterisk 13

[Kumba]

Currently Asterisk 13 is in a final beta release so there still might be issues with compatibility that we haven't found just yet. You also need to make sure your SVN version of ViciDial is new enough that it supports Asterisk 13. The minimum SVN revision for Asterisk 13 support is 2832 but a higher revision is recommended. These instructions should be followed AFTER you have installed and set-up ViciBox phase-2 from the installation manual.


Backup Asterisk Configs (Optional, but recommended)
1) mkdir /root/old-ast
2) cp /etc/asterisk/* /root/old-ast


Remove Asterisk 11 and install Asterisk 13
1) zypper rm asterisk asterisk-dahdi
2) zypper rr 6
3) zypper ar https://download.opensuse.org/repositor ... sk-13.repo
4) zypper --no-gpg-checks ref
5) zypper in asterisk asterisk-dahdi
6) cd /usr/src/astguiclient/trunk
7) perl install.pl --asterisk_version=13 --copy_sample_conf_files --no-prompt
cp /root/old-ast/modules.conf /etc/asterisk/modules.conf (if you backed it up, otherwise copy from another server or don't worry about it)
9) In the web interface, go to Admin --> Servers
10) Click on the server you changed to Asterisk 13
11) Change the asterisk version to '13.18.2-vici' and hit enter to save
12) reboot


And that should be it.

[vkad]

On upgrade manual dialling works, echo test works, but auto dialling is broken. the channels is not being found???

[vkad]

[Jan 31 19:19:20] WARNING[6869][C-0000017e]: func_hangupcause.c:140 hangupcause_read: Unable to find information for channel
_______________________________________________________________________________________________________________
I will try to debug more to see what's wrong ..
---------------------------
VICIDAL 8.0
VERSION: 2.14-650a
BUILD: 180111-1544
Asterisk 13
on a 4 Core 16Gb RAM
___________________

[williamconley]

you should post the lines that lead up to this error, if any. and the carrier config (which determines the channel type).

[frequency]

I did as describe in this and it failed. Only thing i changed was the AST version in server settings to the latest one. When i reload the asterisk, it does not show any channels or anything. When i try to log-in as a user, the call does arrive but "only person in this conference is not there". Tried twice and failed.



Thanks!

[Kumba]

What SVN revision are you on? If you aren't on at least SVN revision 2832 then you do not have Asterisk 13 support.

Also, if you have a G729 codec module make sure you install a new one appropriate for Asterisk 13.

[frequency]

What SVN revision are you on? If you aren't on at least SVN revision 2832 then you do not have Asterisk 13 support.

I am on 2866.

[Kumba]

What's the CLI output when you try to log in?

[frequency]

it says only person in conference but calls wont go through. carrier and phone is peered for sure.

[Kumba]

If you look roughly 8 lines down from when you connected to the Asterisk console you'll see the relevant warning/error message. That error message is "No path to translate from SIP/xyz". This means that the codecs that asterisk has available to it are not compatible with the codecs the phone/carrier is trying to connect with or with the codecs the phone/carrier entry is configured to use.

So basically you have a codec issue. Almost 99% of the time this relates to the G729 codec. Either you need to install one, or the previous codec you installed was for Asterisk 11 and not for Asterisk 13.

You can double check this by making sure the ulaw codec is enabled in the carrier/phone entry you are testing by adding 'allow=ulaw' to it's configuration and removing the 'allow=g729' section.

[frequency]

If you look roughly 8 lines down from when you connected to the Asterisk console you'll see the relevant warning/error message. That error message is "No path to translate from SIP/xyz". This means that the codecs that asterisk has available to it are not compatible with the codecs the phone/carrier is trying to connect with or with the codecs the phone/carrier entry is configured to use.

So basically you have a codec issue. Almost 99% of the time this relates to the G729 codec. Either you need to install one, or the previous codec you installed was for Asterisk 11 and not for Asterisk 13.

You can double check this by making sure the ulaw codec is enabled in the carrier/phone entry you are testing by adding 'allow=ulaw' to it's configuration and removing the 'allow=g729' section.

You were correct, that was due to bad dial plan, allow=g729, added externip to sip.conf and the calls gettings answered in carrier stats but they won't land to agent and asterisk showing weird errors, like the ones we get when manager.conf is bad but if manager.conf is bad, screen's won't show up correctly. Screens are 11.

So i have failed thrice after a fresh install.

[Kumba]

The timeout on non-critical SIP messages also indicate some sort of firewall issue, likely preventing 2-way SIP messages.

[frequency]

The timeout on non-critical SIP messages also indicate some sort of firewall issue, likely preventing 2-way SIP messages.

I always keep the software firewall disabled on the server while the hardware firewall that we have is disabled as well.

[dspaan]

What is the ETA of a new Vicibox version that includes Asterisk 13?

[vkad]

Hi,

When can we expect an asterisk 13 (with opus support) + Viciphone Webrtc iso for vicibox?

Thanks

[Kumba]

I'm already working on ViciBox v.8.1 which ships with Asterisk 13 by default. Probably be done in a week or so.

[williamconley]

I'm already working on ViciBox v.8.1 which ships with Asterisk 13 by default. Probably be done in a week or so.

Could I infer that it will have ViciPhone in it since it has Ast 13?

[vkad]

I'm already working on ViciBox v.8.1 which ships with Asterisk 13 by default. Probably be done in a week or so.

Could I infer that it will have ViciPhone in it since it has Ast 13?

You are a star.

1) Can you please include the sandbox version of asterisk 13 (that includes asterisk amd as no commercial solutions is available for amd as of yet for this asterisk on vici)?
I have tested the sandbox asterisk that has AMD and it runs perfectly fine. Tested with 50 agents dialing 10:1. Used 6 telephony single core 1gb virtual servers and they worked like a charm (with the AMD turned on).
Of course, the other solution would to install additional telephony servers with asterisk 11/ asterisk 13 sandbox to balance the calls with AMD and use the asterisk 13 mainstream for just webrtc.

2) Can you please edit /etc/asterisk/rtp.conf

to either include the stun servers or add the local server ip => public ip in there so that webphone works in NAT.

Just some random observation: Is it just me or asterisk seems to virtualise quite well without any timing problems nowadays. Maybe it was the old hardware previously that did not go quite hand in hand with asterisk virtualisation but now even aws (painfully slow cores) seems to cope well with asterisk virtualisation. Not that I recommend it on a production cluster unless you have a backup dedicated server handy (just in case), but I have not had any problems virtualising vicibox multiple times on various providers as our requirements change from time to time.

Thanks

[dspaan]

I'm confused, with AMD you refer to answering machine detection?

I'm very interested in the status of virtualization as well. I would really love to clone a server for instance and simply the fallback options that it provides. A lot of clients also are worried that the software is run on dedicated servers.

[vkad]

I'm confused, with AMD you refer to answering machine detection?

I'm very interested in the status of virtualization as well. I would really love to clone a server for instance and simply the fallback options that it provides. A lot of clients also are worried that the software is run on dedicated servers.

Yes AMD -> Answering Machine Detection (Even though its not very good, it is still good if you get 70-80% answering machines). Atleast, it suits our purpose.

Virtualisation has not created any time sync, excessive load or other problems for us. We use dedicated server as well and they both run fine. We use VOIP SIP, so we do not need to connect any external hardware on the VMs.

I would be interested to see vicibox as a LXD container. If it would become easy to simply spin up an instance of vicidial, we could potentially see a huge rise in its use. OpenSuse 42.3 does support being run as a container. Containers are as close as you can get to the bare metal performance with the added benefits of snapshots, backups, live migrations and quick reboot times.

[dspaan]

But my problem with AMD is that you get false positives, meaning people who pick up the phone and Asterisk thinks they are answering machines and hangs up.

When running vicidial as a VM for 50 agents do you create just a single server for all these agents? What are the specs looking at memory and processing power?

[vkad]

We have for the VMs
-a single DB server (24gb ddr4, 2.2ghz 8 cores) (handles on average 3000-6000 queries per second when the shift is active)
-a single web (running PHP7 with opcache) + telephony server (agent only server; no dialing allowed; asterisk 13.19.0-vici) (8gb ddr4, 2.2ghz 4 cores)
-5 x telephony servers for balancing calls (the dialer connects to these servers through IAX and places calls) (4gb ddr4, 2.2ghz 2 cores) We use a mix of Asterisk 13 and asterisk 11 here, since these are not connecting to the agents they don't need to be asterisk 11. However, we need the AMD from either ASTERISK 11 or the sandbox version ASTERiSK 13.19.0 so we use a mix for now.

All of the above servers are kept as virtual machines a single E5 v5 + 64gb ddr4. This bare metal has a clone running in High Availability so everything is replicated across the bare metal (we use proxmox so that handles the HA)

So even if the physical machine goes down, we are safe.

[frequency]

We have for the VMs
-a single DB server (24gb ddr4, 2.2ghz 8 cores) (handles on average 3000-6000 queries per second when the shift is active)
-a single web (running PHP7 with opcache) + telephony server (agent only server; no dialing allowed; asterisk 13.19.0-vici) (8gb ddr4, 2.2ghz 4 cores)
-5 x telephony servers for balancing calls (the dialer connects to these servers through IAX and places calls) (4gb ddr4, 2.2ghz 2 cores) We use a mix of Asterisk 13 and asterisk 11 here, since these are not connecting to the agents they don't need to be asterisk 11. However, we need the AMD from either ASTERISK 11 or the sandbox version ASTERiSK 13.19.0 so we use a mix for now.

All of the above servers are kept as virtual machines a single E5 v5 + 64gb ddr4. This bare metal has a clone running in High Availability so everything is replicated across the bare metal (we use proxmox so that handles the HA)

So even if the physical machine goes down, we are safe.

Are you recording calls as well? If yes, are you using an archive server or are they recorded on web server?

[williamconley]

Are you recording calls as well? If yes, are you using an archive server or are they recorded on web server?

I can't say how vkad does it, but I can say that the recordings are MADE on the dialers and then pushed to an archive server that has NO requirements of even being part of the Vicidial cluster. Any FTP/Web server will do.

An FTP server that also has Web is ideal so the files can be pushed via FTP, served via Web and the links can be updated in the Vicidial DB to point to the new location and as of that moment they are no longer a burden on the Vicidial cluster.

But they are made on the dialer unless they are made on an intercept SIP server that's passing the audio through (which is overly complex, but doable).

[frequency]

Are you recording calls as well? If yes, are you using an archive server or are they recorded on web server?

I can't say how vkad does it, but I can say that the recordings are MADE on the dialers and then pushed to an archive server that has NO requirements of even being part of the Vicidial cluster. Any FTP/Web server will do.

An FTP server that also has Web is ideal so the files can be pushed via FTP, served via Web and the links can be updated in the Vicidial DB to point to the new location and as of that moment they are no longer a burden on the Vicidial cluster.

But they are made on the dialer unless they are made on an intercept SIP server that's passing the audio through (which is overly complex, but doable).

Understood! I mean't if he is using his Web Server which is also an "agent only server; no dialing allowed; asterisk 13.19.0-vici". Is this the server that records the calls or is it the other 5 telephony servers that are being used to dial numbers.

[williamconley]

Recording occurs in the Meetme room. The meetme room is on the server that the agent registers to. If the agent registers to multiple servers (load balancing in clusters), then the meetme room is on the dialer that called the agent for This Session.

None of this occurs on a Web server if it is not also a Dialer.

The exception to this is when the recording begins at the inception of the call on a campaign or ingroup before the agent is in the session (possible!), in which case it happens on the server that generated or received the call. Once again, however: Nothing related to the web server unless it also happens to be a dialer capable of receiving or making calls.

[vkad]

I'm already working on ViciBox v.8.1 which ships with Asterisk 13 by default. Probably be done in a week or so.

HI, I hope you are well. Any UPDATES?



1) Can you please include the sandbox version of asterisk 13 (that includes asterisk amd as no commercial solutions is available for amd as of yet for this asterisk on vici)?


I have tested the sandbox asterisk that has AMD and it runs perfectly fine. Tested with 50 agents dialing 10:1. Used 6 telephony single core 1gb virtual servers and they worked like a charm (with the AMD turned on).

Of course, the other solution would to install additional telephony servers with asterisk 11/ asterisk 13 sandbox to balance the calls with AMD and use the asterisk 13 mainstream for just webrtc.

2) Can you please edit /etc/asterisk/rtp.conf

to either include the stun servers or add the local server ip => public ip in there so that webphone works in NAT.

[alo]

We noticed it appears to be working fine on our test system except we notice this message in the asterisk console and we notice a delay of about 10 seconds when logging in before the softphone rings or we press connect to webphone.

Anyone else experiencing this delay?

[Apr 19 17:26:59] WARNING[3306][C-00000000]: func_hangupcause.c:140 hangupcause_read: Unable to find information for channel

[mflorell]

Upgrade to the latest svn/trunk revision, we had to fix the sending script this month to comply with an Asterisk AMI change that was made in the most recent Asteirsk 13 release.

[rrb555]

Same upgraded to 2972 version and still getting the below message after i hungup the call

Code: Select all

WARNING[3192][C-00000004]: func_hangupcause.c:140 hangupcause_read: Unable to find information for channel


I have Asterisk 13.20.0-vici

[mflorell]

That is normal and can be ignored. Not all calls have hangup reasons(like Local/ channels, etc...), but there is no way to only ask for hangup information on certain hung up channels.

[alo]

The SVN upgrade seemed to fix the delay on the webphone connecting. yay!
I am ignoring the Unable to find information for channel. although its spamming the console.

Everything seems to work fine but I am also getting the message every couple seconds:

SRTCP unprotect failed because of authentication failure

Anyone see this happening using webrtc?
Asterisk 13.20.0-vici SVN 2973
Every few seconds

[alo]

I also noticed the time on the bottom of the reports page for the dialing server stops and is updated and re-synced with the database/webserver about every 10 seconds.

while troubleshooting I ran /usr/share/astguiclient/AST_update.pl --quiet on the database/webserver and got the response

pattern match timed-out at /usr/share/astguiclient/AST_update.pl line 470

Line 270 looks to be related to a new manager telnet connection.

This only started happening after upgrading to asterisk 13.

Anything I should be looking at here?

[kashinc]

asterisk 13 has been a nightmare in my cluster as well, none of my servers can stay synched.

[mflorell]

If you upgrade to the latest Asterisk 13.20 version(or higher), you need to make sure you upgrade VICIdial to the latest svn/trunk as well. There were some Manager API changes made in Asterisk that we had to adjust for in the VICIdial code recently.

[kashinc]

I have done both at least 15 times, all the servers still come out of synch for some strange reason... anyone else having this issue?

[williamconley]

pattern match timed-out at /usr/share/astguiclient/AST_update.pl line 470

What's on this line? If it's the telnet login, get the telnet credentials and test a telnet login yourself. Sometimes the host name must be 127.0.0.1 or localhost or the public or private IP. Experiment with it until you get a successful login, then apply your experience to Vicidial and make it work like you did it.

[alo]

So Last night before I gave up for bed I decided to try the perl install.pl --asterisk_version=13 --copy_sample_conf_files --no-prompt again.
And it worked! time stayed in sync. I think out of habbit I may have done --asterisk_version=13.X the first time. not sure if it was a coincidence, but dropping the X fixed the dialing server keeping time. I am still having the problem where agents can't leave a three way call intermittently, but thats a different post. (unless anyone here has a magic solution for it )

Seeing these warnings, but everything seems fine. anything I should be worried about?
SRTCP unprotect failed because of authentication failure
func_hangupcause.c:140 hangupcause_read: Unable to find information for channel
res_agi.c:2262 launch_script: Failed to fork(): Resource temporarily unavailable
pbx.c:7729 pbx_outgoing_attempt: Unable to spawn dialing thread for 'SIP/101'
pbx.c:4637 ast_pbx_start: Failed to create new channel thread

[mflorell]

Also, check your server version in the web admin. If your system is properly configured, you shouldn't be running the "AST_update.pl" script at all under Asterisk 13, it should be running the "AST_update_AMI2.pl" script.

[vkad]

Does the latest version of asterisk have Answering Machine Detection? Also, any chance of a new iso? I was thinking of upgrading to new system over the weekend. any chance that we can get the latest iso before then thanks.