TB 2.8 / Queues / Ring strategies / static agents do not ring

Continuing my rambling, I have seen that when I call a queue and hear its moh (or ringing), the 'queue show' cli command shows:

400 has 1 calls (max unlimited) in 'fewestcalls' strategy (0s holdtime), W:0, C:0, A:1, SL:0.0% within 0s
Members:
Local/299@from-internal/n (Not in use) has taken no calls yet with total talktime 0s
Local/207@from-internal/n (Not in use) has taken no calls yet with total talktime 0s
Callers:
1. SIP/202-b6d50c70 (wait: 0:13, prio: 0)

None of the agents is ringing.

Also, I tried it with dynamic agents and the result was the same.

Same issue here. Callers get MOH and no phones either static or dynamic agents ring. Just get MOH until caller hangs up. Lost....

Hi!

I think this is a problem in the new versions with the order of loading the modules.

If I reload the queue module in the CLI it works. As google told me, that if the queue module laods before the channel modules the agents are "Not in use"

I changed my modules.conf and added the following line as last entry in the [Modules] section:

load => app_queue.so

Now it works.

Regards,

Bernd

A new install of 2.8.0.1 has failed to be able to provide queue functionality for me also.
Is this problem in the trixbox offshoot of freepbx. Would installing proper freepbx fix this. How might this be achieved on a configured system?

Same issue here on a brand new TB install. I've tried with:

preload => chan_local.so
preload => pbx_config.so
and
load => app_queue.so

in modules.conf with no luck (actually the preloads fixed the invalid user you got when running queue show - it now displays Not in use instead for each member)

Can anyone shed some light on this? Would using the 1.4.x asterisk branch fix it? Are you all using 1.6.x?

Same here

From 2.8.0.1 we did a full yum -y update, and now the queues do not work. I have a whole sales pool dead in the water. What gives?!?!?

This appears to be the same issue as shown above. The tail -f /var/log/asterisk/full shows the following:
...

[Dec 10 03:34:51] VERBOSE[11467] logger.c: -- Executing [404@ext-queues:9] Queue("SIP/xxxxx-yyyyyyyy", "404,tr,,custom/SalesQueueMsg,90") in new stack
[Dec 10 03:34:51] NOTICE[11467] app_queue.c: No one is answering queue '404' (10/0/0)
[Dec 10 03:34:55] VERBOSE[11467] logger.c: -- Hold time for 404 is 0 minute(s) 0 seconds
[Dec 10 03:34:55] NOTICE[11467] app_queue.c: No one is answering queue '404' (10/0/0)
[Dec 10 03:34:59] NOTICE[11467] app_queue.c: No one is answering queue '404' (10/0/0)
[Dec 10 03:35:03] NOTICE[11467] app_queue.c: No one is answering queue '404' (10/0/0)
[Dec 10 03:35:07] NOTICE[11467] app_queue.c: No one is answering queue '404' (10/0/0)
[Dec 10 03:35:11] NOTICE[11467] app_queue.c: No one is answering queue '404' (10/0/0)

If there is a live bug in the queues, can someone acknowledge this? Or at least explain where the conflict is?

Is this bug relevant? https://issues.asterisk.org/view.php?id=2574

Or perhaps this bug? https://issues.asterisk.org/view.php?id=16391

And here is yet more info. It seems the db.c is not finding a key. Not sure what this means, but the res_config_mysql fails to connect to the realtime database. In various places, I am hearing this is not an issue. Or is it?

Dec 10 04:21:06] VERBOSE[11687] logger.c: -- Executing [404@ext-queues:10] ESC[1;36;40mSetESC[0;37;40m("ESC[1;35;40mSIP/ddick-b7cfdd00ESC[0;37;40m", "ESC[1;35;40m__CWIGNORE=TRUEESC[0;37;40m") in new stack
[Dec 10 04:21:06] DEBUG[11687] pbx.c: Launching 'Queue'
[Dec 10 04:21:06] VERBOSE[11687] logger.c: -- Executing [404@ext-queues:11] ESC[1;36;40mQueueESC[0;37;40m("ESC[1;35;40mSIP/ddick-b7cfdd00ESC[0;37;40m", "ESC[1;35;40m404,tr,,,90ESC[0;37;40m") in new stack
[Dec 10 04:21:06] DEBUG[11687] app_queue.c: queue: 404, options: tr, url: , announce: , expires: 1260440556, priority: 0
[Dec 10 04:21:06] DEBUG[11687] db.c: Unable to find key '505' in family 'NeoAgenternal'
[Dec 10 04:21:06] DEBUG[11687] db.c: Unable to find key '504' in family 'NeoAgenternal'
[Dec 10 04:21:06] DEBUG[11687] db.c: Unable to find key '503' in family 'NeoAgenternal'
[Dec 10 04:21:06] DEBUG[11687] db.c: Unable to find key '519' in family 'NeoAgenternal'
[Dec 10 04:21:06] DEBUG[11687] db.c: Unable to find key '518' in family 'NeoAgenternal'
[Dec 10 04:21:06] DEBUG[11687] db.c: Unable to find key '517' in family 'NeoAgenternal'
[Dec 10 04:21:06] DEBUG[11687] db.c: Unable to find key '516' in family 'NeoAgenternal'
[Dec 10 04:21:06] DEBUG[11687] db.c: Unable to find key '515' in family 'NeoAgenternal'
[Dec 10 04:21:06] DEBUG[11687] db.c: Unable to find key '514' in family 'NeoAgenternal'
[Dec 10 04:21:06] DEBUG[11687] db.c: Unable to find key '512' in family 'NeoAgenternal'
[Dec 10 04:21:06] DEBUG[11687] app_queue.c: Queue '404' Join, Channel 'SIP/ddick-b7cfdd00', Position '1'
[Dec 10 04:21:06] DEBUG[11687] channel.c: Driver for channel 'SIP/ddick-b7cfdd00' does not support indication 3, emulating it
[Dec 10 04:21:06] DEBUG[11687] channel.c: Set channel SIP/ddick-b7cfdd00 to write format slin
[Dec 10 04:21:06] DEBUG[11687] channel.c: Scheduling timer at 160 sample intervals

I've just switched recently on TB 2.8 and it seems that I have same problem. Further more even on strategy ringall sometimes asterisk doesn't connect to free agents.
I use static agents, but it's really annoying malfunction of PBX.

Regards,
Marcin

I have a clean install of 2.8.0.3 with issues with the Queues that I have now resolved. They do not exactly match these described but are close enough that I thought it worth commenting.

We are using a ringall non-busy strategy with dynamic agents, but once we have more than one agent logged in these refuse to ring.

Closer examination suggested that the dialparties.agi script never finishes. Having looked at the php and added various debug comments I discovered that the is_ext_avail fails at the point it makes its manager call (is this due to overlapping log-ins or a manager bug??).

Anyway, I have temporarily fixed this by moving the two lines that instantiate the manager php object and connect it to the server into the is_ext_avail function. This may seem inefficient, but in our circumstances this is the only place it was used and it is only used once, hence it shouldn't be that bad. (oh and eventually I realised I need to edit /var/www/html/admin/modules/core/agi-bin/dialparties.agi so that it doesn't get overwritten!!!).

Anyway, this is working, suggesting to me that the cause of these issues may be the manager object used by dialparties.agi failing to work correctly, having been created as the script first fired. Opinions and longer-term fixes from anyone who actually knows the internals would be very welcome.

I have a customer with the same issue.
He is running TB 2.8.0.3.
Works fine on RINGALL but not with ROUNDROBIN

I have done a good amount of searching with regards to this issue. I do not see any mention of it outside of Trixbox. This concerns me as all indications seem to point towards Asterisk as the issue. Some have mentioned that the issue did not occur until updating to the recent releases of Asterisk.

I have been using Trixbox for a number of years and am in the process of performing an upgrade to V2.8.0.3. However, this is a big issue for us. I am not able to operator the queues as we have done for years. I also do not see any solutions. I have attempted to investigate the changes to the modules.conf. First off app_queue.so is not in the file in the first place. Second reloading and unloading/loading the app_queue.so does not fix the problem, so I have little confidence that any changes to the modules.conf are going to resolve this issue.

With the second proposed solution I am not sure if it is a type or if the user is possible on a different version of Asterisk, but where he has:

$astman = new AGI_AsteriskManager( );
if (!$astman->connect("127.0.0.1", $ampmgruser , $ampmgrpass)) {
return 0;
}

I have:

$astman = $AGI->new_AsteriskManager();
if (!$astman->connect("127.0.0.1", $ampmgruser , $ampmgrpass)) {
exit (1);
}
}

Even though there is a difference I attempted to make the changes suggested but it did not seem to make a difference. Although I moved the following as it would be the complete statement and not just the suggested suggested.

if (!$has_extension_state) {
$astman = $AGI->new_AsteriskManager();
if (!$astman->connect("127.0.0.1", $ampmgruser , $ampmgrpass)) {
exit (1);
}
}

Does anybody know what the issue is? Is there a solution to this problem? Why is there no mention of this issue with other systems?

I would like to get to the bottom of this. As I mentioned before this seems like a pretty big issue.

Regards,
-Eugene

I can confirm I am having the exact same issue.

Versions:

Trixbox v2.8.0.3
Asterisk 1.6.0.10-FONCORE-r40

Queue Status:

700 has 0 calls (max unlimited) in 'leastrecent' strategy (0s holdtime), W:0, C:0, A:0, SL:0.0% within 0s
Members:
Local/200@from-internal/n (dynamic) (Not in use) has taken no calls yet with total talktime 0s
Local/201@from-internal/n (dynamic) (Not in use) has taken no calls yet with total talktime 0s
No Callers

DEBUG Output:

[Jan 29 12:50:57] DEBUG[13293]: app_queue.c:3573 try_calling: SIP/192.168.2.45-b6b40128 is trying to call a queue member.
[Jan 29 12:50:57] DEBUG[13293]: db.c:197 ast_db_get: Unable to find key '201' in family 'NeoAgenternal'
[Jan 29 12:50:57] DEBUG[13293]: app_queue.c:2484 ring_one: Skipping 'Local/201@from-internal/n' with metric 0 because it is not online
[Jan 29 12:50:57] DEBUG[13293]: db.c:197 ast_db_get: Unable to find key '200' in family 'NeoAgenternal'
[Jan 29 12:50:57] DEBUG[13293]: app_queue.c:2484 ring_one: Skipping 'Local/200@from-internal/n' with metric 0 because it is not online
[Jan 29 12:50:57] DEBUG[13293]: app_queue.c:2462 ring_one: Nobody left to try ringing in queue
[Jan 29 12:50:57] NOTICE[13293]: app_queue.c:2699 wait_for_answer: No one is answering queue '700' (2/0/0)

I have no solution yet, I will carry on searching for the problem.

Hi! :

First excuse my poor English =).
Im facing the same problem, and after read the previous posts, I notice this line:

[Dec 10 04:21:06] DEBUG[11687] db.c: Unable to find key '512' in family 'NeoAgenternal'

I have the same lines:

[Jan 30 09:43:43] DEBUG[23782] app_queue.c: SIP/502-08cc3398 is trying to call a queue member.
[Jan 30 09:43:43] DEBUG[23782] db.c: Unable to find key '503' in family 'NeoAgenternal'
[Jan 30 09:43:43] DEBUG[23782] app_queue.c: Skipping 'Local/503@from-internal/n' with metric 0 because it is not online
[Jan 30 09:43:43] DEBUG[23782] db.c: Unable to find key '501' in family 'NeoAgenternal'

then added manually in the cli this keys:

trixbox1*CLI> database put NeoAgenternal 501 TRUE
Updated database successfully
trixbox1*CLI> database put NeoAgenternal 503 TRUE
Updated database successfully
trixbox1*CLI>
501 and 503 are the agent numbers

then make some tests and the queue start working. Both agents dynamic and static seem to work properly after adding the keys.
The problem is, when restarting the server or the asterisk or the agent logout then you need to add the keys again,

Still doing tests, and looking for some real solution but maybe this can help in the meantime.

Regards

----
Edgar
La Paz - Bolivia

I feel the pain as I spent about three days smacking my head on the table as well. I am glad to see that m_michael50 may have a possible perminate solution. I do want to thank garuso for your post. If I do not hear from m_michael50 I may have to go that route to get things going.

I do have to say that I am surpprised that this has not been a bigger issue and addressed sooner...

-Eugene

I can now confirm that in my test system the queues problem is also solved when I compile Asterisk 1.6.0.21.

(Un)fortunately, I do not feel competent enough to do this operation on my productive system. Do we have any idea when TB will incorporate this *isk release?

Thanks

Hi,

I tried the new update of TB with Asterisk 1.6.0.22 today.

Sadly, the queues did not work again.

I did two tests: a) one clean install with just a queue and two extensions and b) restore my original config from my working installation where I also have the queues problem.

In both cases, the queues did not ring any agents unless I set them to "ringall".

Given that I (and Michael above) was succesful when I compiled 1.6.0.21 on my test system, could this mean that the queues are broken *again* one asterisk version later?

Since it seemed really odd to have the problem solved in one version (1.6.0.21) and re-broken in the next (1.6.0.22), I did the following test:

-Installed TB 2.8.0.3 plain vanilla. Did NOT yum update. - Queues not working
-Compiled Asterisk 1.6.0.21 on top of this installation without anything else - Queues worked
-Updated TB via yum update to *isk 1.6.0.22 (among all other things) - Queues did not work. Even after going to PBX settings and re-saving my test extensions and queues.
-Downloaded and compiled 1.6.0.22 on top of this 'updated' installation (was already 1.6.0.22 through yum)-Queues worked!

I guess this goes to show that it is not an Asterisk problem but rather a problem of how Trixbox implements Asterisk in the distribution.

Note: By PlainVanilla I mean an empty TB install with just two sip extensions and one queue. Absolutely no other configuration.

Any other thoughts?

Hi Henry,

Thanks for the walkthrough above.

What I failed to comprehend is why you compile asterisk 1.6.0.21 on top of the TB/Ast 1.6.0.22 rpm. Why not compile the 1.6.0.22 asterisk? Or any other for that matter. Is there a specific reason you want to stick with 1.6.0.21?

BR

Henry, can you share new app_queue.so for us?

Me too.... Just posted another thread about it:

http://www.trixbox.org/forums/trixbox-forums/help/queue-strategy-...

How can Trixbox have a distribution specific problem that so many people are reporting, and leave it go unfixed for this long?!

Clearly from the descriptions of being able to compile asterisk from native source and drop in app_queue.so over the top of it that points squarely at a problem with Trixbox, not an upstream issue?

I have been installing a trixbox system to replace an old (like ancient old) asterisk install and happen to run across this gem of an issue.

After looking at bbhenry's solution I was somewhat apprehensive to do it as the app_queue application in the trixbox distribution has quite a few changes that I didn't want to lose. Looking at the code the problem turned out to be a partial implementation of something called NeoAgents (google wasn't that helpful but didn't really matter). The implementation is only partial however, and broken at that, in that it builds a string buffer with corrupted information (hence the NeoAgenternal in garuso's post above). So I disabled this weird check as it's pretty much not doing anything that I can tell. Below is the patch diff for those interested.

Another thing that might cause issues for some people if you are using standard extensions as static members is that the none of the extensions will get called after an asterisk restart until you reload the queue. This is due to the queue module being loaded before the dial plan and can be resolved by adding the lines below to your /etc/asterisk/modules.conf beneat the [modules] section.

load => pbx_config.so
load => chan_local.so

As it can be quite a pain to compile this update here's a link to an app_queue.so to use if you'd like. This was compiled for my own use and so the usual disclaimers apply. It was compiled against asterisk16-1.6.0.22 so you will most likely need that version on your machine.
http://www.megaupload.com/?d=OYU54U2U

Diff for app_queue:

diff -uNr asterisk16-1.6.0.22/apps/app_queue.c asterisk16-1.6.0.22-fixqueue/apps/app_queue.c
--- asterisk16-1.6.0.22/apps/app_queue.c        2010-02-12 15:35:29.000000000 -0500
+++ asterisk16-1.6.0.22-fixqueue/apps/app_queue.c       2010-03-19 15:20:02.000000000 -0400
@@ -764,10 +764,12 @@
                                result = QUEUE_NO_UNPAUSED_REACHABLE_MEMBERS;
                        } else {
 
+                               /*
                                result = get_member_status_from_db(member);
 
                                if(result != QUEUE_NORMAL)
                                        continue;
+                               */
 
                                ao2_unlock(q);
                                ao2_ref(member, -1);
@@ -2550,11 +2552,14 @@
                        }
                } else {
                        /* Agent with name including '#30' is linked agent*/
+                       /*
                        if(strstr(best->interface, "#30\0") || (QUEUE_NORMAL == get_member_status_from_db(best->member)))
                        {
+                       */
                                /* Ring just the best channel */
                                ast_debug(1, "Trying '%s' with metric %d\n", best->interface, best->metric);
                                ret = ring_entry(qe, best, busies);
+                       /*
                        }
                        else
                        {
@@ -2562,6 +2567,7 @@
                                best->chan = NULL;
                                best->stillgoing = 0;
                        }
+                       */
                }
        }

Yes, simply copy it into the modules folder (keeping a backup of the original of course) and restart the asterisk to test it out. The only difference between it and the stock trixbox module is that I've disabled the incomplete and broken code that prevents the ring methods from working. You'll probably need to have the same version of asterisk as I listed in the post as well.

I found that after update asterisk, not all modules was updated, cdr did not workig etc.
so step 3 should be:
yum update asterisk16\*

Chameleoncoder - I tried your module and it does in fact allow the different ring strategies to work. But if one of the agents in the queue is on the phone and that is the agent that asterisk tries to dial (based on ring strategy), the caller gets a 'That extension is busy' message. Is the result = get_member_status_from_db(member); line that got commented out the piece that figures out which agents are actually available? It is possible that I'm doing something wrong, but in testing this is what I hit up against. Thanks for your efforts.

Tom

It's possible that is what is occurring. Snip from cli:

Members:
Local/2050@from-internal/n (Not in use) has taken no calls yet with total talktime 0s
Local/2051@from-internal/n (Not in use) has taken no calls yet with total talktime 0s
No Callers

These were setup via the Trixbox web gui (same queue show output is produced when logging in to the queue as a dynamic agent from an endpoint). Are you saying the line should read SIP/2050@from-internal instead? How would this be done differently?

Thanks again!

Tom

Thanks for your help. Hopefully this will get fixed soon. FYI to anybody interested - we have another server running Trixbox 2.8.0.1 (Asterisk 1.6.0.9-samy-r27) where the queues work properly.

bump.

01-Jun-2010!!

Just did a yum update, still broken.
Asterisk 1.6.0.22-samy-r60 built by root @ revisor.trixbox.com

This issue is open since Sep-09, soon a whole year have passed.

Does Fonaility support Trixbox CE and its community or is it dead?

Update to asterisk 1.6.0.26 and execute this on asterisk console for each agent

database put NeoAgenternal 335 TRUE

335 is the agent number

This is my version and my work around:
Asterisk 1.6.0.26-FONCORE-r78 built by root @ revisor.trixbox.com on a i686 running Linux on 2010-06-08 22:01:27 UTC
Trixbox: 2.8.0.4-1

The "database put" seems to work with dynamic and static clients.

database put NeoAgenternal 335 TRUE

And that seems to stick through a reboot so here is a fix for a fresh install so when adding a new extension you can have the entry automatically added.


--- /var/www/html/admin/modules/core/functions.inc.php 2010-06-30 20:18:36.000000000 -0600
+++ /var/www/html/admin/modules/core/functions.inc.php 2010-06-30 20:15:36.000000000 -0600
@@ -2694,6 +2694,7 @@
$existingdevices = $astman->database_get("AMPUSER",$user."/device");
if (empty($existingdevices)) {
$astman->database_put("AMPUSER",$user."/device",$id);
+ $astman->database_put("NeoAgenternal",$user,"TRUE");
} else {
$existingdevices_array = explode('&',$existingdevices);
if (!in_array($id, $existingdevices_array)) {

Hopefull this helps in the automation of adding extensions.