Re: correlating ASH and OSWatcher data

From: Ryan January <>
Date: Thu, 20 Feb 2014 11:48:19 -0600
Message-ID: <>

Hi Adric,
Do you feel the server processes were errantly consuming CPU outside of a session or are you trying to identify your top CPU consumers in that time frame?
My first step is generally adding up CPU consumption per session by summing the TM_DELTA_CPU_TIME or the count of 'ON CPU' session state samples for the desired time frame. Generally your outliers will surface quickly. Unfortunately I've not yet found a rock solid OS PID to session mapping using only ASH/AWR. I have had some mild success with using info found in the "program" column of the ASH table, using the process identifier (eg: LGWR or LMS1) and pull the pids from logs on the file system.

On 02/20/2014 11:01 AM, Adric Norris wrote:
> I'm in the process of troubleshooting a recent performance issue
> (currently bypassed), which occurred as a result of a database server
> becoming overloaded from a CPU perspective. At this point we've
> identified a number of OS dedicated server processes which seem to
> have been driving excessive load, via the OSWatcher data, and we also
> have a number of suspect database sessions identified via ASH.
> So my question is this... is there a way to tie the 2 sources
> together? It doesn't appear that the necessary [g]v$process
> information is captured by ASH, so I suspect this isn't possible, but
> it's certainly possible that I've missed a necessary detail.
> The database in question is 64-bit, running on a Solaris 10
> SPARC 64-bit server.
> Thanx!
> --
> "I'm too sexy for my code." -Awk Sed Fred


This email is intended solely for the use of the addressee and may
contain information that is confidential, proprietary, or both.
If you receive this email in error please immediately notify the
sender and delete the email..

Received on Thu Feb 20 2014 - 18:48:19 CET

Original text of this message