Elastix Clustering, !!! Extensions problem

jad.jabali

Joined
Jun 8, 2010
Messages
6
Likes
0
Points
0
#1
Hi

I've just finish setup two elastix Boxes, as Clustering using DRBD and Heartbeat.

Every thing is working just perfect, any changes done in the Primary get replicated to the secondary and Vice-versa.

Till the moment i tried to make a call between two Extensions, and i got "The number you've dialed is not in service" and the same for 7777

This how it is look like:

-- Executing [7777@from-internal:1] Goto("SIP/105-00000039", "from-pstn|7777|1") in new stack
-- Goto (from-pstn,7777,1)
-- Executing [7777@from-pstn:1] Set("SIP/105-00000039", "__FROM_DID=7777") in new stack
-- Executing [7777@from-pstn:2] NoOp("SIP/105-00000039", "Received an unknown call with DID set to 7777") in new stack
-- Executing [7777@from-pstn:3] Goto("SIP/105-00000039", "s|a2") in new stack
-- Goto (from-pstn,s,2)
-- Executing [s@from-pstn:2] Answer("SIP/105-00000039", "") in new stack
-- Executing [s@from-pstn:3] Wait("SIP/105-00000039", "2") in new stack
-- Executing [s@from-pstn:4] Playback("SIP/105-00000039", "ss-noservice") in new stack
-- <SIP/105-00000039> Playing 'ss-noservice' (language 'en')
== Spawn extension (from-pstn, s, 4) exited non-zero on 'SIP/105-00000039'
-- Executing [h@from-pstn:1] Hangup("SIP/105-00000039", "") in new stack
== Spawn extension (from-pstn, h, 1) exited non-zero on 'SIP/105-00000039'

I don't know why asterisk assume that calls are coming form "from-pstn"

Each box has two Ethernet interfaces and floating IP address when it is active.

Any Idea Guys !!!
 

Megabyte

Joined
Mar 28, 2009
Messages
327
Likes
0
Points
0
#2
Hi, you have to know what extensions are reserved for freepbx, in feature codes the 7777 is reserved for Simulate Incoming Call, go to your features code tab and verify the extensions you cant you for your internals extensions.
 

danardf

Joined
Dec 3, 2007
Messages
8,069
Likes
10
Points
88
#3
Hi all.

jad.jabali, if you have a process to do it, could you send me it?
I'm interested to do it, and compare it with the others ways.

Thanks ;)
 

jad.jabali

Joined
Jun 8, 2010
Messages
6
Likes
0
Points
0
#4
Hi Megabyte

Sorry for being not clear, i meant even when I'm trying to simulate the call dialing 7777 I've got the same message.
 

Megabyte

Joined
Mar 28, 2009
Messages
327
Likes
0
Points
0
#5
Hi, you need to create an incoming route to this DID or you will need to do and incoming route with any DID/Any CID. It looks like you dont have a route.
 

jad.jabali

Joined
Jun 8, 2010
Messages
6
Likes
0
Points
0
#6
Hi

I've read through many tutorials, give me some time and i will try to combine a clear guide
how to do it and how to fix all the possible problems that could face you in the proccess...
 

danardf

Joined
Dec 3, 2007
Messages
8,069
Likes
10
Points
88
#7
OK It's cool.
Thanks

Best regards
 

jad.jabali

Joined
Jun 8, 2010
Messages
6
Likes
0
Points
0

danardf

Joined
Dec 3, 2007
Messages
8,069
Likes
10
Points
88
#9
Thanks very much :)

Mega cool!
 

dicko

Joined
Oct 24, 2008
Messages
4,099
Likes
0
Points
0
#10
Thanks jad.jabali :

That's a nice distillation.

If you don't mind, I will add a couple of my own two cents worth, which I find will actually make it work in a "real world" Elastix type deployment.

First off, I found that drbd83 is a little more robust that drbd, both for the server and the kmod

other services likely to running on your boxen also need supporting, both the data structures will need similarly replicating and the services need disabling and adding to the /etc/ha.d/haresources file

If you use the dhcp server and/or tftpd provisioning, you will need to also do the same for that server
(/var/lib/dhcpd/* and /tftpboot )

If any sort of faxing is involved then hylafax also (that includes avantfax)
hylafax service and
/var/spool/hylafax/

You will also have a problem (which might or might not bug you, it does me) with iaxmodem when it is not running and the faxgettys being very noisy in the logs. you can migrate control from inittab to runsv as the supervisor for the faxgettys (runsvdir-start in inittab and comment out the faxgettys)

If you are using Aastra XML then comment out the daemons on rc.local and use this half -assed script I wrote for the /etc/init.d directory to control the aastra daemons:-

astra:
----------------------------------------------------------------------------------
#!/bin/bash
RETVAL=0

case "$1" in
start)
if [ ! `ps ux | awk '/aastra_daemon1/ && !/awk/ {print $2}'` ] ;then echo starting aastra1;( /usr/bin/php -q /var/www/html/aastra/asterisk/aastra_daemon1& ); fi

if [ ! `ps ux | awk '/aastra_daemon2/ && !/awk/ {print $2}'` ] ;then echo starting aastra2;( /usr/bin/php -q /var/www/html/aastra/asterisk/aastra_daemon2& ); fi
;;

stop)
killall /usr/bin/php -q /var/www/html/aastra/asterisk/aastra_daemon*
;;
*)
echo $"Usage: $0 {start|stop}"
exit 1
esac

exit $RETVAL


---------------------------------------------------------------------------------

There might well be others that you use like postfix, messaging, a2billing etc. but that depends on your particular deployment, but the methodology is the same, just "expand to include"

If you have a public IP that is available for the server, I suggest you don't use a "floating" IP for security reasons (why expose three servers when you only need to expose one) If you need help with that I can possibly further expound as the gateway and ip addresses need "special attention" and you will need a backdoor into those replicated servers, I also find that replicating over a VLAN is more easily controlled, if you use redfone stuff, take their advice and dedicate that interface for that hardware (192.168.1.0/24 by default and to save yourself later problems, resolved probably by fonulator2 ) and add a VLAN on that eth with QOS which can isolate the sync data without impinging on that layer 2 stuff, or even consider using a null-modem serial link :)


dicko

p.s.

My methedolgy which I find all-in-all is easier, is to build a "onesided" server with all the services and patches you need, test it, deploy it, do the rsync/tar/symlink etc. into the drbd partition, (test, deploy and use again , rigourously!) if everything is ok a week later (see I'm a coward) then clone it with mondoarchive (with or without the replicated data, depending on the "clone" hardware if it;s the same its easier, if it's not you will have to be careful). Then edit the new clone's identity (names and IP's) as appropriate, delete /var/lib/heartbeat/hb_* on the erstwhile secondary (which will save a whole shit-load of problems if you clone) make sure the primary is functional and then bring up the secondary and let it sync. Obviously you can actually start with a functional server and do all this with no "downtime" (if you get it right :) ) (ok , on a production server you might need a temporary usb disk for a while, but it's possible.)

JM2CWAE (there is no better feeling as a sysop than knowing that when a server blows up (which WILL happen) , you can think to yourself, "Who the fuck cares, I'll fix it next Tuesday")
 

danardf

Joined
Dec 3, 2007
Messages
8,069
Likes
10
Points
88
#11
Hey Dicko. How are you my friend!! :woohoo:

Good work for these explanations no?
 

dicko

Joined
Oct 24, 2008
Messages
4,099
Likes
0
Points
0
#12
Franck, salut!

I am well, et tu?

Yes, he did a good job of writing it all down in one place,
I'm personally trying to migrate from heartbeat to pacemaker but there are a lot of dependencies that need resolving with the current Elastix kernel, I will perhaps wait for a stable 2.0

Regards to you and your famille.


dicko
 

danardf

Joined
Dec 3, 2007
Messages
8,069
Likes
10
Points
88
#13
I'm well thanks.

I think that the Elastix 2.0 seems not really ready to be stable. It should be waiting 1 month more. Maybe August!
It's my mind, now.... I don't know. :huh:
 

Members online

No members online now.

Latest posts

Forum statistics

Threads
30,901
Messages
130,885
Members
17,561
Latest member
marouen
Top