Re: AWR Sample Report

On Dec 6, 4:47 am, raja <dextersu..._at_gmail.com> wrote:
> Hi Steve,
>
> I took the output of the query which you have given in the previous
> post.
> I am not able to post the output file here. So i am sending you and
> charles.
> The file contains the output that was taken from 25/11/2008 -
> 03/12/2008
> In average, it looks like ordering the output by total waits, shows
> "PX qref latch" as the priority.
> I searched regarding this and found that it is related to parallel
> processing.
> ( I also feel that they might have used parallel hint to make the
> queries faster. Let me check out on these. )
> Please check the file that i have sent to ur mail id and provide me ur
> suggestions.
>
> With Regards,
> Raja.

Raja,

I examined the spreadsheet - negative numbers should not appear when subtracting the old values from the new values. You may need to make an adjustment to the SQL statement if you plan to look at more than one wait event at a time, or if there are more than one database instances.
The SQL statement might need to be modified like this (I do not have an AWR license, so this might not work as written): select
  to_char(begin_interval_time,'YYYY_MM_DD HH24:MI'),   dhse.instance_number,
  time_waited_micro - nvl(lag(time_waited_micro, 1, 0) over (partition by dhse.instance_number, dhse.event_name order by dhse.snap_id),0) as time_waited,
  total_waits - nvl(lag(total_waits, 1, 0) over (partition by dhse.instance_number, dhse.event_name order by dhse.snap_id),0) as total_waits
from
  dba_hist_snapshot dhs,
  dba_hist_system_event dhse
where
  dhs.snap_id = dhse.snap_id
  and dhs.instance_number = dhse.instance_number order by
  1,
  2;

Just out of curiosity, I would like to see an updated version of the spreadsheet using the new SQL statement.

In short, the "PX qref latch" wait event may not be the problem.

Charles Hooper
IT Manager/Oracle DBA
K&M Machine-Fabricating, Inc. Received on Mon Dec 08 2008 - 08:56:03 CST