RE: ECOM170 system on the blink

Is this normal?

 

 

 

-----Original Message-----
From: midrange-l-bounces@xxxxxxxxxxxx
[mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of Crump, Mike
Sent: Wednesday, June 21, 2006 12:56 PM
To: Midrange Systems Technical Discussion
Subject: RE: ECOM170 system on the blink

 

FYI - run priority and time slice is normally set by the job class.  In

a normal world you would be dealing with the QBATCH class and the QBATCH

sbs.  You might want to confirm those settings.

 

-----Original Message-----

From: midrange-l-bounces@xxxxxxxxxxxx

[mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of Dwayne Allison

Sent: Wednesday, June 21, 2006 9:33 AM

To: Midrange Systems Technical Discussion

Subject: RE: ECOM170 system on the blink

 

Hello,

 

This is what our operator says:

 

Two things are happening that I have noticed. 1-some batch jobs are

coming out to execute at a RUNPTY 10 and high time slice. 2 - We are

running performance collection.  I am trying to change job priority as I

see them running.  The only thing on the system that runs at a priority

10 is DSP01.  If anything runs equal to or above that we could possibly

have a situation where the DSP01 could be non-responsive.

 

-----Original Message-----

From: midrange-l-bounces@xxxxxxxxxxxx

[mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of Crump, Mike

Sent: Wednesday, June 21, 2006 8:13 AM

To: Midrange Systems Technical Discussion

Subject: RE: ECOM170 system on the blink

 

You might check the job logs of those jobs that ended just to see what

they might indicate.  It could be a failed cache as mentioned else where

or a rogue job or situation.  I'd check on those items I mentioned

before and see if you see anything out of the ordinary.  WRKACTJOB is

not the most accurate but if you don't have WRKSYSSTS (comes with

performance tools) it will have to do.  Open it, run it for at least a

minute, sort it by CPU utilization and see what you get.  On DSPSYSSTS

look for high fault rates or any wait-inel numbers.  On WRKDSKSTS look

for high disk utilization - one disk or many....again, not knowing what

you have normally makes it hard to say what would be bad.  In my world

all of my drives operate under a 15% utilization.  Just don't take a 5

second snapshot - let it run for at least 60 seconds or more.

 

-----Original Message-----

From: midrange-l-bounces@xxxxxxxxxxxx

[mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of Dwayne Allison

Sent: Wednesday, June 21, 2006 8:57 AM

To: Midrange Systems Technical Discussion

Subject: RE: ECOM170 system on the blink

 

 

Can you clarify about the system kicking you off?  

1) I did a DBU on a file and it tried to open the file and after five

minutes of trying to open, my session ended.  I had to log back on again

2) I was coding a program in one session and went to another session to

check a spool file.  I put a 5 on the spool file and after about 3

minutes of trying to open the spool file my session ended (both of them)

3) I was trying to look at a 2 page spool file and it took about 5

minutes before it opened.

 

 

-----Original Message-----

From:

midrange-l-bounces+dwayne.allison=affirmativeinsurance.com@xxxxxxxxxxxx

[mailto:midrange-l-bounces+dwayne.allison=affirmativeinsurance.com@midra

nge.com] On Behalf Of Crump, Mike

Sent: Wednesday, June 21, 2006 7:39 AM

To: Midrange Systems Technical Discussion

Subject: RE: ECOM170 system on the blink

 

Dwayne,

 

Can you clarify about the system kicking you off?  I am assuming you

mean that while signed on you session might hang or you get kicked back

to the signon screen?

 

Most of the time poor system performance comes to very hot cpu, memory

demand problems, or disk performance.  Sometimes these can be affected

by a single bad job.  As your system is in a slow period I would check

the CPU utilization (either through WRKACTJOB or WRKSYSACT), memory

demand via DSPSYSSTS, and disk utilization via WRKDSKSTS.  Are you

running any performance data collection during this time frame?

 

CPU - look for jobs taking a high level of CPU%, especially those jobs

with a high run priority like an interactive job.

 

Memory - check your page fault numbers.

 

Disk - look at your disk drive utilization.  See if it is high - this

could even just be one drive.  I don't know what is normal for you so

you will have to interpret the results yourself.  

 

If you are having interactive jobs ending I would inspect your job log

for any information and possibly also check the QSYSOPR message queue

for anything that might look suspicious.

 

Also, when you indicate slow I am assuming that interactive response

get's sluggish?  Batch jobs seem to be running ok?

 

-----Original Message-----

From: midrange-l-bounces@xxxxxxxxxxxx

[mailto:midrange-l-bounces@xxxxxxxxxxxx] On Behalf Of Dwayne Allison

Sent: Wednesday, June 21, 2006 7:51 AM

To: Midrange Systems Technical Discussion

Subject: ECOM170 system on the blink

 

Good Morning,

 

 

 

 

 

Our ECOM170 box has become slow at time and it sometimes kicks us off

the system.  We check the system status and we are at 77 %.