[Asterisk-Users] app_queue, fewestcalls and leastrecent logic

[Jim Friedeck]

In our original spec for Digium, leastrecent was specifically 'agent who 
answered a call longest ago for that queue'. (not a direct quote) It 
would then go to the next agent in order of 'longest go'. Has this 
changed? Does it immediatly go roundrobin by agent number or agent load 
order? Thanks.

Jim Friedeck

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

Brian West wrote:

>Ok just had my boss point something out:
>
>"I'd think dumping calls on most-idle would be fairly straightforward, but
>could be skewed if agentA is on a 40 minute call, agentB has a bunch of 5
>minute calls"
>
>So total call time should be counted in the logic somewhere.
>
>bkw
>
>On Sun, 10 Aug 2003, Brian West wrote:
>
>  
>
>>I think we are starting to see what type of logic people are wanting in
>>fewestcalls and leastrecent strategy.
>>
>>bkw
>>
>>On Sun, 10 Aug 2003, Richard Lyman wrote:
>>
>>    
>>
>>>i disagree, instead of thinking 'fallback' how about 'order' the agents
>>>(by effecting the 'metric') so you 'target' the agent you want first
>>>then if fail they go right to the next one in the 'ordered' list.
>>>
>>>Brian West wrote:
>>>
>>>      
>>>
>>>>leastrecent suffers the same fait as fewestcalls onlying ringing the
>>>>leastrecent agent over and over endlessly.  It should have a fallback
>>>>option.
>>>>
>>>>roundrobin with leastrecent first
>>>>roundrobin with fewestcalls first
>>>>
>>>>I would like to see a roundrobin with leastbusy first option.
>>>>(just because you have taken less call or leastrecent doesn't mean you
>>>>haven't been a busy agent!)
>>>>
>>>>I'm sure better autologoff logic as per my first email would be a great
>>>>idea also.
>>>>
>>>>bkw
>>>>
>>>>On Sun, 10 Aug 2003, Richard Lyman wrote:
>>>>
>>>>
>>>>
>>>>        
>>>>
>>>>>well if you ask me, the leastrecent part would work if you reversed the
>>>>>logic on the metric.
>>>>>
>>>>>my other last_used mod would do a time_t on that agent the last time it
>>>>>was 'tried' (ast_request'd) then (i was using arrays) qsort so that (new
>>>>>agents) '0' would be on top, and the agent that got the most recent
>>>>>attempt would be on the bottom '1057174447' (below is an example)
>>>>>
>>>>>   -- sorted agent array: 317 last_used: 0
>>>>>   -- sorted agent array: 318 last_used: 0
>>>>>   -- sorted agent array: 319 last_used: 0
>>>>>   -- sorted agent array: 300 last_used: 1057174447
>>>>>
>>>>>that way, (for leastrecent anyway), you are always working with a full stack of agents.
>>>>>
>>>>>
>>>>>
>>>>>Brian West wrote:
>>>>>
>>>>>
>>>>>
>>>>>          
>>>>>
>>>>>>First of all I would like to thank Mark for getting roundrobin to go
>>>>>>roundrobin.  Good job.
>>>>>>
>>>>>>Now we have some options here for leastrecent and fewestcalls strategy. It
>>>>>>needs some work on the logic and Mark recommend that I ask the list and
>>>>>>get some input before he makes any changes to it.
>>>>>>
>>>>>>fewestcalls from what I have seen would always ring the agent with the
>>>>>>fewestcalls first then go into roundrobin if that agent didn't answer.
>>>>>>
>>>>>>Next new caller would ring fewestcalls agent first then start roundrobin.
>>>>>>
>>>>>>What do you think should happen in fewestcalls?  Right now it just rings
>>>>>>the agent with the fewestcalls over and over with current app_queue logic.
>>>>>>
>>>>>>leastrecent from what I have been looking at will ring the agent that has
>>>>>>least recently take a call first then if they don't answer go into
>>>>>>roundrobin.  Then the next new call coming from queue would first go to
>>>>>>the leastrecent first then try every agent in roundrobin till answered
>>>>>>then starting over again.  New caller from queue hits leastrecent agent
>>>>>>first.
>>>>>>
>>>>>>Same thing happens in leastrecent strategy. The leastrecent agent will
>>>>>>ring over and over with current app_queue logic.
>>>>>>
>>>>>>Now some of you might recommend autologoff options.  But that also might
>>>>>>need some work.  I don't want to log off an agent for not answering the
>>>>>>phone only once.  So here is how I would like to see autologoff work.
>>>>>>
>>>>>>Example:
>>>>>>queue timeout = 20
>>>>>>agent autologoff = 60
>>>>>>
>>>>>>The agent would have to not answer their phone 3 times in a row to get
>>>>>>logged off.  As it stands now they did not answer just once and get logged
>>>>>>off.  Thus allow for an employee to use the excuse for not working when
>>>>>>they should be logged in and taking calls.
>>>>>>
>>>>>>Unless i'm wrong here.
>>>>>>
>>>>>>Please post your input on these options and how you would like them to see
>>>>>>them function function.
>>>>>>
>>>>>>Thanks,
>>>>>>Brian
>>>>>>CWIS Internet Services
>>>>>>
>>>>>>
>>>>>>_______________________________________________
>>>>>>Asterisk-Users mailing list
>>>>>>Asterisk-Users at lists.digium.com
>>>>>>http://lists.digium.com/mailman/listinfo/asterisk-users
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>            
>>>>>>
>>>>>_______________________________________________
>>>>>Asterisk-Users mailing list
>>>>>Asterisk-Users at lists.digium.com
>>>>>http://lists.digium.com/mailman/listinfo/asterisk-users
>>>>>
>>>>>
>>>>>
>>>>>          
>>>>>
>>>>_______________________________________________
>>>>Asterisk-Users mailing list
>>>>Asterisk-Users at lists.digium.com
>>>>http://lists.digium.com/mailman/listinfo/asterisk-users
>>>>
>>>>
>>>>
>>>>        
>>>>
>>>_______________________________________________
>>>Asterisk-Users mailing list
>>>Asterisk-Users at lists.digium.com
>>>http://lists.digium.com/mailman/listinfo/asterisk-users
>>>
>>>      
>>>
>>_______________________________________________
>>Asterisk-Users mailing list
>>Asterisk-Users at lists.digium.com
>>http://lists.digium.com/mailman/listinfo/asterisk-users
>>
>>    
>>
>_______________________________________________
>Asterisk-Users mailing list
>Asterisk-Users at lists.digium.com
>http://lists.digium.com/mailman/listinfo/asterisk-users
>
>
>  
>