asterisk users - Apr 2010 - [PATCH] Make Queue announcements more consistent (1.4.26.2)

Hi,
After playing around with queues a bunch on 1.4.26.2, I noticed a few things,
which the patch below addresses. It addresses:
- Callers in position 0 will hear periodic/position announcements at a
very different rate than all other callers.
    -- Announcements while in position 0 could be delayed up to
"timeout+retry" seconds.
    -- This patch reduces that possible delay to only "timeout"
seconds
- The say_position and periodic_announcement times are in elapsed time
that _includes_ the
time of the announcement.
    -- This patch changes those times to be the time _between_ playing
of those announcements

Thanks.

-- James


--- asterisk-1.4.26.2/apps/app_queue.c	2009-08-10 13:14:34.000000000 -0700
+++ asterisk-1.4.26.2.new/apps/app_queue.c	2010-04-25 22:25:08.000000000 -0700
@@ -345,6 +345,7 @@
 	time_t last_periodic_announce_time; /*!< The last time we played a
periodic announcement */
 	int last_periodic_announce_sound;   /*!< The last periodic
announcement we made */
 	time_t last_pos;                    /*!< Last time we told the user
their position */
+	time_t last_ring_time;              /*!< Last time we tried to ring
the agents */
 	int opos;                           /*!< Where we started in the queue */
 	int handled;                        /*!< Whether our call was handled */
 	int pending;                        /*!< Non-zero if we are
attempting to call a member */
@@ -1653,6 +1654,7 @@
 		res = 0;

 	/* Set our last_pos indicators */
+	time(&now);
 	qe->last_pos = now;
 	qe->last_pos_said = qe->pos;

@@ -2131,6 +2133,8 @@
 	if (!res)
 		ast_moh_start(qe->chan, qe->moh, NULL);

+	/* Refresh now so that frequency is time _between_ recordings */
+	time(&now);
 	/* update last_periodic_announce_time */
 	qe->last_periodic_announce_time = now;

@@ -3292,7 +3296,8 @@
 static int wait_a_bit(struct queue_ent *qe)
 {
 	/* Don't need to hold the lock while we setup the outgoing calls */
-	int retrywait = qe->parent->retry * 1000;
+	//int retrywait = qe->parent->retry * 1000;
+	int retrywait = RECHECK * 1000;

 	int res = ast_waitfordigit(qe->chan, retrywait);
 	if (res > 0 && !valid_exit(qe, res))
@@ -4003,6 +4008,7 @@
 	qe.max_penalty = max_penalty;
 	qe.last_pos_said = 0;
 	qe.last_pos = 0;
+	qe.last_ring_time = 0;
 	qe.last_periodic_announce_time = time(NULL);
 	qe.last_periodic_announce_sound = 0;
 	qe.valid_digits = 0;
@@ -4074,9 +4080,12 @@
 				break;
 			}
 			/* Try calling all queue members for 'timeout' seconds */
-			res = try_calling(&qe, args.options, args.announceoverride,
args.url, &tries, &noption, args.agi);
-			if (res)
-				goto stop;
+			if ((time(NULL) - qe.last_ring_time) > qe.parent->retry) {
+				res = try_calling(&qe, args.options, args.announceoverride,
args.url, &tries, &noption, args.agi);
+				qe.last_ring_time = time(NULL);
+				if (res)
+					goto stop;
+			}

 			stat = get_member_status(qe.parent, qe.max_penalty);

@@ -4125,7 +4134,7 @@
 			/* If using dynamic realtime members, we should regenerate the
member list for this queue */
 			update_realtime_members(qe.parent);

-			/* OK, we didn't get anybody; wait for 'retry' seconds; may get
a
digit to exit with */
+			/* OK, we didn't get anybody; poll our retry */
 			res = wait_a_bit(&qe);
 			if (res)
 				goto stop;

On 27/04/10 2:21 AM, James Lamanna wrote:
> Hi,
> After playing around with queues a bunch on 1.4.26.2, I noticed a few
things,
> which the patch below addresses. It addresses:
> - Callers in position 0 will hear periodic/position announcements at a
> very different rate than all other callers.
>      -- Announcements while in position 0 could be delayed up to
> "timeout+retry" seconds.
>      -- This patch reduces that possible delay to only "timeout"
seconds
> - The say_position and periodic_announcement times are in elapsed time
> that _includes_ the
> time of the announcement.
>      -- This patch changes those times to be the time _between_ playing
> of those announcements
Please post this to issues.asterisk.org.

Unfortunately developers are unable to look at or add patches without 
knowing the license.

When you create an account on issues.asterisk.org you can file a 
disclaimer for the code and then the patch can be added to the base 
Asterisk install (assuming it meets coding guidelines etc).

-- 
Cheers,

Matt Riddell
Managing Director
_______________________________________________

http://www.venturevoip.com/news.php (Daily Asterisk News)
http://www.venturevoip.com/exchange.php (Full ITSP Solution)
http://www.venturevoip.com/st.php (SmoothTorque Predictive Dialer)

Hi All,

I am using Asterisk as my pbx talking to a main proxy server.
The main Proxy server is sending unsolicited Notify Messages to the clients
after a call is established.

Is there a setting that I can tell Astersik to forward any NTY received from
Proxy to be forwarded to the End users?

2) Also Can I have the translation from User at domain to User at domain. (
right Now I am having dail plan as numbers)
Now I want to support soft clients also whose User names are not numbers only
but also Alpha numeric character,

I am looking something like:
bob at asterisk.com to be mapped to bob at external.com  what is the config
changes :-)

TIA.


      
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
http://lists.digium.com/pipermail/asterisk-users/attachments/20100426/3fa2aa3c/attachment.htm