Locking of 'app_queue'

We are currently running with 15 agents on a system with significant call volumes (relative to the number of agents). After about 15-20 minutes of a good run, the app_queue module locks up.

Does anyone have any idea on how to obtain greater stability or fix this problem all together? As we must restart Asterisk to get it running again, which of course is not a good thing.

This is the problem we are having with v1.2.12.1 still:


Looks like a fix was added to v1.2.12.1, but the comments at the end indicate this patch was also required:

nic_bellamy - reporter
08-17-06 16:22

I've just uploaded [b]manager_eventq_backport-1.2.10.patch[/b], a backport of the trunk producer-consumer changes to the manager eventq.

I’m not sure if it fixes this bug, but it’s worth a go, and I’d be interested to hear other peoples results.[/quote]

Anyone have any idea where this patch was ‘uploaded’ to?

I can’t comment on your second question, muppetmaster, but as for the instability with AgentCallBackLogin, several developers have stated that it’s being deprecated in 1.4 and won’t exist in 1.6 - AddQueueMember and RemoveQueueMember will most likely be replacing AgentCallBackLogin.

FWIW, we have been using AQM/RQM in production across 8 servers for almost a year now (mostly 1.2.4 boxes) without any issues, except for some bugs with devicestate and the queue sending multiple calls to agents (these can be worked around by using the Local channel and the groupcount functions).

I’m sure you have probably seen these, but here is a way to emulate AgentCallBackLogin functionality with AQM:

voip-info.org/wiki/view/Aste … nformation
voip-info.org/wiki/view/Agen … nt+channel

Anyways, I know this doesn’t really answer your question, but maybe it will give you some options, at least…

Yes Muppet, Wes is correct. We also moved to dynamic members becuase of queue locking issues and it has been running very smooth since.

Thank you for the feedback, but I need to focus on fixing the current functionality due to timeframes, and not least because it is reported to be fixed. Looking at the Backports page:

asterisk-backports.org/wiki/inde … .2#Patches

It appears others are looking for this mysterious location that this patch has been posted to:

[quote=“Backports”]Lots of people have asked for a backport of the new Manager non-blocking event queue as the 1.0 and 1.2 blocking model is known to cause deadlocks in a variety of interesting ways. Unfortunately none of us know how to do it.(kebl0155)

* UPDATE: I've just uploaded a backport of this to 1.2.10 to [bugs.digium.com/view.php?id=6626](http://bugs.digium.com/view.php?id=6626) (nicb, Fri, 18 Aug 2006 09:33:34 +1200)[/quote]

muppetmaster - look at the last patch uploaded for that bug - it appears to be the backport you are referring to. it’s for 1.2.10, but i’m assuming you can probably backport it further if needs be…

direct link to patch: bugs.digium.com/file_download.ph … 3&type=bug