Re: How to handle NGS-IQ member full messages?

On 5/25/2017 5:15 PM, John Yeung wrote:

We have a "power user" who does a lot of queries, and we get CPA5305
("Record not added. Member QIQ_PCD is full. (C I 9999)") messages
often enough that it's annoying and disruptive.

What are our options for handling this so that the queries run but
don't need manual intervention to complete?

If this were a permanent PF, my inclination would be to adjust its
size, but it's generated on the fly in QTEMP.

There's no 'per user' defaults for creating tables, so you're stuck with
system-wide solutions like ADDRPYLE or changing the default for CRTPF
(or whatever command ultimately does that chore). That said, there are
options that are system wide which can be focussed to individual users.
You may need to really crank up the job logging or even fire up
performance tools and trace an NGS job in order to work out what exactly
is creating the file.

1) Monitor QSYSOPR with a program that answers only CPA5305 and only for
that user and only for n times (in case of a runaway). The rest of the
messages get re-sent to whomever actually monitors them.

2) The QIBM_QCA_CHG_COMMAND exit point can let you alter the CRTCMD on
the fly for this user so that you can in fact 'change the file' as it's
being created.

3) In the old days I used to stuff my own CRTPF command higher in the
library list than IBM's, and in my version I'd write the code to impose
my will upon the user :-) These days, many vendors are hard-qualifying
their system command calls to force the use of IBM's base commands. I
think the current best practise is to create a proxy command. Still, it
might work for NGS.