Date:         Mon, 27 Oct 1997 19:22:06 -0500
Reply-To:     CICS List <CICS-L@UGA.CC.UGA.EDU>
Sender:       CICS List <CICS-L@UGA.CC.UGA.EDU>
From:         Christopher Frank <FAdvice@AOL.COM>
Subject:      Re: MAXKEYLENGTH in LSR pools

Larry, Tara,
 
Two different approaches are being discussed here. MAXKEYLENGTH(255) insures
that the maximum will be handled and will consume additional virtual storage
for placeholder blocks (PLHs). However, it will not delay startup.
Additionally, if files are dynamically added later you know they will work.
 
Omitting MAXKEYLENGTH or specifying MAXKEYLENGTH(0) forces CICS to issue
SHOWCAT calls to DFP for each file to obtain the keylength in order to
compute this value. This insures that PLHs are not overallocated, but slows
startup (significantly if any of the datasets have been archived) and
prevents you from dynamically adding new files to the LSRPOOL with keys
larger than CICS calculated when the pool was created.
 
Which of the above is more important depends on your installation. As PLHs
are above the line now, I would opt for MAXKEYLENGTH(255).
 
BTW, the message you are seeing was added recently by PN92737, to warn that a
significant delay could occur if the dataset(s) CICS issues SHOWCATs against
(to obtain the information) have been migrated by HSM. Nothing has changed;
this has always been true. It's just a warning to explain why a delay has
occurred, should one be encountered.
 
PQ03510 was released to correct a problem with the above PTF. The warning
message is issued needlessly  in certain cases; you may want to obtain that
PTF.
 
Regards,
 
Christopher Frank
STAR, Inc.