Hi sandro, ran into your query/ can you make some sense of these values?
(3600*24) delta
and
having round(max(delta),0) > 300

select
sql_id,
count(*),
round(min(delta),0) mn,
round(avg(delta),0) av,
round(max(delta),0) mx,
substr(max(times),12) max_run_time
from (
select
sql_id,
sql_exec_id,
max(delta) delta ,
— lpad sets a fixed width on delta so it can be
— stripped off above with substr
— delta in “times” is just for sorting not displaying
lpad(round(max(delta),0),10) || ‘ ‘ ||
to_char(min(start_time),’YY-MM-DD HH24:MI:SS’) || ‘ ‘ ||
to_char(max(end_time),’YY-MM-DD HH24:MI:SS’) times
from ( select
sql_id,
sql_exec_id,
cast(sample_time as date) end_time,
cast(sql_exec_start as date) start_time,
((cast(sample_time as date)) -
(cast(sql_exec_start as date))) * (3600*24) delta
from
dba_hist_active_sess_history
where sql_exec_id is not null
)
group by sql_id,sql_exec_id
)
group by sql_id
having round(max(delta),0) > 300
order by mx
/

select
     SQL_ID ,
     sum(decode(session_state,'ON CPU',1,0))     "CPU",
     sum(decode(session_state,'WAITING',1,0))    -
     sum(decode(session_state,'WAITING', decode(wait_class, 'User I/O',1,0),0))    "WAIT" ,
     sum(decode(session_state,'WAITING', decode(wait_class, 'User I/O',1,0),0))    "IO" ,
     sum(decode(session_state,'ON CPU',1,1))     "TOTAL"
from   dba_hist_active_sess_history 
where SQL_ID is not NULL  
 and  dbid=&dbid
   -- and sample_time between 
   --     to_date('07-NOV-2011 16:00','DD-MON-YYYY HH24:MI') and
   --     to_date('07-NOV-2011 17:00','DD-MON-YYYY HH24:MI') 
group by sql_id
order by sum(decode(session_state,'ON CPU',1,1))   desc
/

col av for 9999999
col mx for 9999999
col mn for 9999999
 
select
   sql_id,
   count(*),
   round(min(delta),0) mn,
   round(avg(delta),0) av,
   round(max(delta),0) mx,
   substr(max(times),12) max_run_time
from (
   select
        sql_id,
        sql_exec_id,
        max(delta) delta ,
        -- lpad sets a fixed width on delta so it can be
        -- stripped off above with substr
        -- delta in "times" is just for sorting not displaying
        lpad(round(max(delta),0),10) || ' ' ||
        to_char(min(start_time),'YY-MM-DD HH24:MI:SS')  || ' ' ||
        to_char(max(end_time),'YY-MM-DD HH24:MI:SS')  times
   from ( select
                                            sql_id,
                                            sql_exec_id,
              cast(sample_time as date)     end_time,
              cast(sql_exec_start as date)  start_time,
              ((cast(sample_time    as date)) -
               (cast(sql_exec_start as date))) * (3600*24) delta
 
           from
              dba_hist_active_sess_history
           where sql_exec_id is not null
        )
   group by sql_id,sql_exec_id
 
)
group by sql_id
-- incorrectly was
-- having mx > 300
-- this is the right syntax
having round(max(delta),0) > 300
order by mx
/

Good questions. One is easier than the other

1) Why did I use 0 for the min bucket? It was the easiest quickest thing to put for minimum value. A better value would have been the min value ever seen for that query as the starting bucket. I thought about this but never got around to it.

2) Why did I pass in mxdelta+.1 for the max bucket? Well this is a bit of hack that I didn’t look into closely but at least meant to document but forgot to.
With width_bucket, if I ask for 5 buckets I get 2 extra for a total of 7.
The bottom is anything below my minimum, which in this case will always have 0 values as no query should have a negative response time, so I can ignore this bucket. The largest will be anything starting with my largest value on up so if I add a tiny bit to my max value, then this bucket never have values and always be 0 so I can ignore it and I’m back to my 5 buckets.
What if I just asked for 4 buckets instead? Wouldn’t this work? This would give me 6 buckets and I could ignore the first giving me 5 buckets, right?
If I asked for 4 buckets then the fifth bucket would only have single max values in it , ie no range. For the most part this bucket will just have the value 1, unless the exact maximum elapsed time happened more than once. I’d prefer the 5th bucket to actually show a range rather than a single value.