Barbara Nitz wrote on 01/25/2006 01:40:23 AM:
> >IPCS won't necessarily think about the WHOLE data space, just enough
> >locations adjacent to the final byte in the space to fill the screen on
> >the storage panel. Smart code processing data spaces notices such things
> >as invalid segments in a data space and produces big numeric answers
> >along the way. Similar smart code processing address spaces also thinks
> >about invalid regions.
> So IPCS is even smarter than I thought!
It has to be. Run the numbers, and going byte by byte or page by page
would produce response times that you could not survive a typical debugging
session.
>
> Whenever I had similar performance problems, my dump directory was very
much
> fragmented with CI/CA splits. It usually helped to just delete/define the
> dump directory and redo the analysis. If you still have that problem with
a
> fresh directory (you could use a different name to start with and repro
the
> old data in - just make it big enough), then at least you'll know it's
not
> VSAM related. What I also do in cases like this is take a dump of my hung
> user and look at the trace table to see what I am doing.
I periodically discipline myself to go through my inventory, throw out the
descriptions of things no longer of interest, and then run the following
CLIST that must be 25 years old:
PROC 1 DSN
CONTROL LIST
ALLOCATE FILE(SAVEDDIR) UNIT(SYSVIO) CYL SPACE(10 10) +
RECFM(U) BLKSIZE(3072) REUSE NEW
REPRO INDATASET(&DSN) OUTFILE(SAVEDDIR)
REPRO INFILE(SAVEDDIR) OUTDATASET(&DSN) REUSE
FREE FILE(SAVEDDIR)
It should run a little faster than REPROs between two directories because
only the 2nd REPRO pass builds an index.
Bob Wright - MVS Service Aids
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO
Search the archives at http://bama.ua.edu/archives/ibm-main.html
