Re: Long running backups to a 3590-B11

We only have 12 GB, running at about 75% used - it has been in excess of 
100%, but normally fluctuates between 69% & 78% - we recently had a scenario 
which we labeled lemon/400 - our hardware reps told us we were getting an 
upgrade, but in reality they installed a downgrade.  Speed of processor & 
amount of memory can have astronomical impact.

News/400 has a book on Mastering/400 Performance

How often do you re-IPL to reclaim disk space associated with humongous print 
jobs?  We get about 5% disk space back when we do a re-IPL.

Review GO CLEANUP settings.

We have had several upgrades in which some types of backups went faster & 
some went slower ... we used to do GO BACKUP daily on just our most active 
libraries & GO SAVE biweekly for a SAVE 21 because GO BACKUP took 90 minutes 
while SAVE 21 took 6 hours, but now we do SAVE 21 every nite because it takes 
90 minues while the GO BACKUP takes 3 hours.

My theory is that GO BACKUP is now doing some kind of start stop tape 
movement on each library, where previously it was a smoother copy process, 
and that this is related to AS/400 model, processor, and OS/4 version.

Since we converted from MAPICS to BPCS, and installed UPS for key 
communications hardware, we have had to do one restore from backups about 
once every 3-5 months, while in the olden MAPICS days without UPS, we had to 
restore from backups several times a week.

If your need to restore is comparable then not saving access paths makes 
sense.

WRKF look for QHST* in QSYS ... you may only need a day or two activity - 
adjust System Value QHSTJOBLOG

WRKPRB work with problems & see if there are unneccessary problem logs ... 
System Value QPRBHLDITV governs how long they are kept ... last time I tried 
to muck with this it would not let me lower it below 30 days

I got some code off one of the trade zine forums to DSPFD all files all 
libraries to a temporary area, that should be in QTEMP, then Query off it to 
show how much disk space is wasted by which files that contain records coded 
for deletion in need of reorg ... when we started this, some files had 
millions of such records, but now the report runs off the job scheduler 
regularly, and I do a reorg when the wasted space exceeds 10 meg.

>  From:    BullJ1@Midas-Kapiti.com (Bull, Jeff)

>  Has anyone else had a similar problem -
>  
>  I have a system with 102GB of DASD, running at about 80% used.  It has
>  regularly taken between 11 to 13 hours to do a full system backup each
>  weekend (SAVSYS + NONSYS + SAVDLO + SAV (ifs)).  All of a sudden, the
>  backups started taking 15/16 hours.  I have recently upgraded to v4r3 (from
>  v4r1), initiallly the backups jumped to 28 hours!!  IBM recommended we 
apply
>  additional PTFs, a series of group-ptf packages, this cut the 28 to 18 but
>  this is still very painful to system availablility, as you can imagine.
>  
>  I am liaising with IBM and with the CE; I've stopped saving access paths,
>  I'm getting staff to delete old/redundant data, what more can I do?  Has
>  anyone had similar experiences or have some advice ? 
>  
>  Kind regards,
>  Jeff Bull.
>  Senior AS/400 Support Consultant (AS/400)
>  Midas Kapiti International Ltd
>  


Al Macintyre  ©¿©
http://www.cen-elec.com MIS Manager Programmer & Computer Janitor
http://www.whma.org = our nitch industry has a large committee searching for 
ErPdMes products right for WHM without work-arounds, which for many members 
resemble a snipe hunt - if you are in this software business & want to fix 
your product so it will not be mistaken for what your competition has been 
doing, then you might want to join our great game.

Y2K is not the end of my universe, but a re-boot of that old Chinese curse.
The road to success is always under construction.
Stiff in opinions, always in the wrong.
When you want it cheap - you get what you paid for.
When in doubt, read the manual, assuming you can find the right one.
+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to MIDRANGE-L@midrange.com.
| To subscribe to this list send email to MIDRANGE-L-SUB@midrange.com.
| To unsubscribe from this list send email to MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---