HP3000-L Archives

October 1999, Week 4

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Carl McNamee <[log in to unmask]>
Reply To:
Carl McNamee <[log in to unmask]>
Date:
Thu, 28 Oct 1999 16:35:49 -0500
Content-Type:
text/plain
Parts/Attachments:
text/plain (149 lines)
A good tool for checking what a program is doing is SPT from HP.  It will
help you pinpoint inefficiencies in programs.  It has helped us solve
several "performance problems" like you described and prevented others from
going into production.  The cost is not extravagant either.

Carl McNamee
Systems Administrator
Billing Concepts

-----Original Message-----
From: L. A. Barnes [mailto:[log in to unmask]]
Sent: Thursday, October 28, 1999 4:14 PM
To: [log in to unmask]
Subject: Re: JSMAIN.PUB.SYS


HP (and other hardware vendors) love it when you add more hardware to solve
a
problem.  At my last job that was the buzz word, until a great DBA (TR) did
some
investigating into the programs that were running.  T.R. was able to put a
plan
together that revamped some of these programs and the systems performance
improved
drastically.

Memory can be a viable solution, but when you mentioned this huge COBOL
program
that was modified maybe the problem is still in the program.  It's always
worth
investigating, unless money is not an issue.



Rick Clark wrote:

> David,
>
> I understand that this is not much info, and I guess I am feeling more
> like L.A.Barnes' little guy banging my head against the computer
> today...... It has been one of those dazes!
>
> After further research, it turns out that a few more lines of code were
> added to a huge COBOL program that when installed, the system takes a
> nose dive. When removed, the system operates normal. We are thinking
> that this additional code, when multiplied by 500+ users stored in
> memory, caused the HP3000 performace bottlenecks. It is my understanding
> that the 3000 systems, once it reaches a certain 'boiling' point, will
> drop in performance. Not a gradual curve, but a drop.
>
> Bottom line is, we are buying more memory. This I hope will resolve our
> problem.
> I'm not convinced this is the problem, but it doesn't hurt. What I am
> questioing now is that the Configuration guide (1995) shows the 959ks
> has a max of 2048 memory. We were told that it is capable of 8gb and
> that MPE/iX has a limit of 3.75gb. Is this true? What is the true max
> amount of memory I can install in the system?
>
> Thanks again in advance.
>
> Rick Clark
> Senior Systems Analyst
> WW&R
> Cleveland, Ohio
>
> David Gale wrote:
> >
> > Rick,
> >
> > unfortunately system performance has many more variables and I am afraid
the
> > information you have provided doesn't provide an imediate answer. JSMAIN
> > does not appear to be a resource hog at his point.
> >
> > I will say that in the past when a sudden change of performance was
noticed,
> > and there didn't seem to be a change in anything else, the first thing I
> > looked at was disc space. Make sure you are not using more than 85% on
your
> > volume sets.
> >
> > To really get a good look at performance issues, you need a good tool.
<plug
> > alert> I favor the Lund Performance Tools myself <end-plug>. You really
need
> > something that not only tells you the current pulse points of your
system
> > but allows you to capture trending information.
> >
> > There could be any number of problems therefore I cannot with honesty
give a
> > better opinion of your problem.
> >
> > Good Luck!
> >
> > Dave Gale
> >
> > > -----Original Message-----
> > > From: Rick Clark [mailto:[log in to unmask]]
> > > Sent: Thursday, October 28, 1999 10:26 AM
> > > To: [log in to unmask]
> > > Subject: JSMAIN.PUB.SYS
> > >
> > >
> > > What is this program? We are running extremely slow the past few days
> > > and nothing has changed that could have caused this (at least that is
> > > what they are claiming). I am looking at the SHOWQ;ACTIVE and noticing
> > > in the B queue one PIN consistently. This turns out to be
> > > JSMAIN.PUB.SYS
> > >
> > >
> > > Q  PIN   JOBNUM                           Q  PIN   JOBNUM
> > >
> > >                                           B   856
> > >                                           C  M2652 #S1449
> > >                                           C  U1263 #S1642
> > >                                           C  U1108 #S1531
> > >                                           C  U1452 #S1599
> > >                                           C  U2146 #S1442
> > >                                           C  U2709 #S837
> > >                                           C  U2747 #S1626
> > >                                           C  U2883 #S1459
> > >                                           D  U186  #J1221
> > >                                           D  U2547 #J1116
> > >
> > >
> > >                     ------QUANTUM-------
> > > QUEUE  BASE  LIMIT  MIN    MAX    ACTUAL  BOOST  TIMESLICE
> > > -----  ----  -----  ---    ---    ------  -----  ---------
> > >  CQ    152    200   1      2000   68      DECAY    200
> > >  DQ    202    238   2000   2000   2000    DECAY    200
> > >  EQ    240    253   2000   2000   2000    DECAY    200
> > >
> > > PUB SYS:DO SHOWP
> > > PUB SYS:SHOWPROC ;PIN=856;SYSTEM
> > > QPRI  CPUTIME   STATE  JOBNUM  PIN  (PROGRAM) STEP
> > >
> > > B100* 0:00.010  WAIT           856  (JSMAIN.PUB.SYS)
> > > PUB SYS:
> > >
> > > Thanks in advance....
> > >
> > >
> > > Rick Clark
> > > Senior Systems Analyst
> > > WW&R
> > > Cleveland, Ohio
> > >

ATOM RSS1 RSS2