Thanks Rob and Barton.  I'm betting that I can attribute this then to
the process "t" which is started by cron often.  It seems to grow bit by
bit until the next recycle and then starts over.  It's being phased out,
so I'm not going to sweat it :)

Let us know if you need some pathological cases.  We're good at that..


Marcy Cortes


"This message may contain confidential and/or privileged information.
If you are not the addressee or authorized to receive this for the
addressee, you must not use, copy, disclose, or take any action based on
this message or any information herein.  If you have received this
message in error, please advise the sender immediately by reply e-mail
and delete this message.  Thank you for your cooperation."


-----Original Message-----
From: Linux on 390 Port [mailto:[EMAIL PROTECTED] On Behalf Of
Barton Robinson
Sent: Monday, August 28, 2006 10:45 PM
To: [email protected]
Subject: Re: [LINUX-390] Init process consuming CPU

The totals is a sum of all processes - the ESALNXP report shows
processes above the threshold of .1%.  So quite possibly, there are many
http servers or java servers using less than .1%.  You would NOT want a
report that showed all the processes.

The other report is the ESALNXA report. This report shows CPU by
application. So, for example in this report, all the processes started
by process 17860 or 15625 (the parents that spawned these httpd servers)
would be accumulated and reportedcu - but a lot of that doesn't show up
when you just look at processes using more than .1%.

There is soon also to be a ESALNXU report that shows useage by logon
userid - in answer to many requests.

To further Rob's response. Any process that dies during an interval has
their current interval CPU added to their parent.
If the parent happens to be init, then "init" cpu grows in the
"children".  This shows up here.  For example, cron starts up, uses some
CPU, stops.  This cpu gets allocated to init children.
Applications don't usually have this scenario.

I expect capture ratio currently of processes to be real close to 100%.
I expect the '*totals' to be very accurate as well.  Rob and others have
been providing pathalogical cases to test this - and still rarely find a
way to show room for improvement....


>Date:         Mon, 28 Aug 2006 18:20:07 -0500
>Reply-To: Linux on 390 Port <[email protected]>
>From: Marcy Cortes <[EMAIL PROTECTED]>
>
>It's consistently 6 and 7% of an IFL on 1 http server.  On it's twin on

>the other LPAR, it's consistently about 1.5%.  So, is there that much 
>"leftover" because of the httpd processes?  We were wondering about the

>differences - and also why the velocity numbers don't add up - is that 
>a
>bug?:
>
>Here's the "twin"
>                            <-Process Ident-> <-----CPU
Percents----->=20
>Time     Node     Name      ID    PPID   GRP   Tot  sys user syst
usrt=20
>-------- -------- --------- ----- ----- ----- ---- ---- ---- ----
----=20
>18:18:00 LNXC9136 httpd     17876 17860  2454  0.1  0.1  0.0  0.0
0.0=20
>                  httpd     17872 17860  2454  0.1  0.1  0.0  0.0
0.0=20
>                  httpd     15651 15625  2454  0.1  0.1  0.0  0.0
0.0=20
>                  httpd     15650 15625  2454  0.1  0.0  0.0  0.0
0.0=20
>                  httpd      8063  8057  2454  0.1  0.0  0.0  0.0
0.0=20
>                  httpd      7926  7916  2454  0.1  0.1  0.0  0.0
0.0=20
>                  t          2642  2637  2624  0.1  0.0  0.1  0.0
0.0=20
>                  t          2632  2625  2624  0.2  0.0  0.2  0.0
0.0=20
>                  rotatelo   2457  2454  2454  0.2  0.2  0.0  0.0
0.0=20
>                  snmpd      2323     1  2322  0.7  0.5  0.2  0.0
0.0=20
>                  init          1     0     0  1.6  0.0  0.0  1.2
0.4=20
>                  *Totals*      0     0     0  5.9  2.1  2.0  1.2
0.5=20
>
>
>
>
>Marcy Cortes







"If you can't measure it, I'm Just NOT interested!"(tm)

/************************************************************/
Barton Robinson - CBW     Internet: [EMAIL PROTECTED]
Velocity Software, Inc    Mailing Address:
 196-D Castro Street       P.O. Box 390640
 Mountain View, CA 94041   Mountain View, CA 94039-0640

VM Performance Hotline:   650-964-8867
Fax: 650-964-9012         Web Page:  WWW.VELOCITY-SOFTWARE.COM
/************************************************************/

----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions, send
email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or
visit http://www.marist.edu/htbin/wlvindex?LINUX-390

----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390

Reply via email to