[asterisk-users] Stable-Stable Asterisk

[Steve Totaro]

Joshua Colp wrote:
> Doug Lytle wrote:
>   
>> Tzafrir Cohen wrote:
>>     
>>> On Fri, Aug 24, 2007 at 07:33:21AM -0400, Steve Totaro wrote:
>>>   
>>> stability problems) with 1.0, ahve already migrated to 1.2 or 1.4, and
>>> now swear (by?) 1.2 or 1.4.
>>>
>>>   
>>>       
>> My decision based on what I've been reading in the bug tracker and 
>> people commenting on how they've had to roll back to 1.2 to regain a 
>> stable system.  We are not having issues with our 1.2.x installs, but 
>> I've been 'encouraged' by the development team to upgrade to 1.4.
>>     
>
> I'll just chime in for those who are thinking of moving to 1.4 and do 
> end up having issues... don't just turn around and go back to 1.2 
> immediately. File a bug report with all the needed information so things 
> can get fixed. As a development team we can't test every single scenario 
> possible with Asterisk, we depend on the users to tell us if there are 
> problems and tell us how to reproduce them. Asterisk only gets better 
> thanks to the users out there. If you file a bug report keep on top of 
> it... if more information is needed, provide it. I've had a few bugs 
> where the reporter dropped off the radar and I had to end up trying 
> every possible configuration combination to find the bug and fix it, 
> taking away time that I could have spent on other issues.
>
> I'm going to end this email with a question myself... how many people 
> have Asterisk on a development/staging server before deployment, test, 
> and isolate the issues they may have in their specific scenario?
>
>   

I always do.  A dev system with a a few calls here and there.  Checkout 
the cool new features, syntaxes, and nuances.

Here is a better question.

How can I setup a Dev system to handle 15,000 calls a day using nine 
queues, 200 agents, nine servers, seven quad port T1 Sangoma boards 
connected to a Adtran DS3 MUX?  Now add specific scenarios such as AGI 
(of whatever form) manager connections controlling your CRM, screenpop, 
sales processes, and reporting.  You cannot scale that up on a dev system.

Now imagine if the system is down you lose $26,000/hr.  You know what, I 
am going back to 1.2.x as quick as possible, then I will try to assemble 
some data and open a bug, but good luck with that. 

This exact same scenario played out on 1.2.  It worked flawlessly until 
the manager flaked out under REAL load.

Thanks,
Steve

Dev systems are great, so are small PBXs for companies that get a few 
dozen calls a day.