50002 errors can never be eliminated entirely, but how do you know you have done enough to eliminate them? What is normal vs excessive? We have developed a metric that can be shared across Cisco partners and customers that will help us all determine where we stand in the industry.
Run the query below against your UCCE instance and share the results. This query basically provides statistics and ultimately a ratio of 50002 errors per agent hour logged in (50002dedups_Hours_%).
select
convert(date, Date) Date,
count(1) Agents,
1.0*nullif(sum([50002_dedups]), 0) / nullif(sum(HoursLoggedIn), 0) [50002dedups_Hours_%],
sum(Logins) Logins,
sum(HoursLoggedIn) HoursLoggedIn,
sum([50002_dedups]) [50002_dedup],
sum([50002s]) [50002s]
from
(
select
*,
datediff(HH, LoginDate, LogoutDate) HoursLoggedIn -- hours logged in
from (
select
SkillTargetID,
convert(date, DateTime) Date,
max(case when Event=1 then 1 else 0 end) Logins, -- only count 1 login per agent per day
min(case when Event=1 then aed.DateTime else null end) LoginDate,
max(case when Event=2 then aed.DateTime else null end) LogoutDate,
sum(case when Event=3 and ReasonCode=50002 then 1 else 0 end) [50002s],
sum(case when Event=3 and ReasonCode=50002 and Duration<>900 and Duration<>1800 then 1 else 0 end) [50002_dedups] -- remove duplicate 50002 (15 min interval = 900 seconds, 30 min interval = 1800 seconds)
from Agent_Event_Detail aed
where aed.DateTime >= convert(date, getdate()-14) -- last 14 days
group by aed.SkillTargetID, convert(date, DateTime) -- group by agent, then day
) A
) B
group by convert(date, Date)
order by convert(date, Date)
I get asked about this error code everywhere I go, and when it is not correlated to a server-side or network event, it is often a rat hole that never seems to end. Use the results of this query to help determine if it is worth the effort or not to investigate causes of 50002 in your environment. The power of this data is when you share your results with the community. Thanks for participating.
