Re: SQL stored procedures and activation groups via JDBC

Mark S. Waterbury wrote:

lloen wrote:

...(snip)...
What are the symptoms? Long term slowdown, increased paging, or simply more and more disk consumed?

What I would predict is that you would see the storage growth for each of these jobs take place to a certain point over time and then simply "top off" and not grow any more.

Whether this did more than simply waste a bit of disk space would depend on other factors.

The crux of the issue seems to be, is there a way to configure QZDADSOINIT jbos to run in an ILE activation group? How can they
do this?

For example, should they just put ALL of the i5/OS *PGMs and
*SRVPGMs into a single large named activation group, by
specifying e.g. ACTGRP(QILE) on every *PGM and *SRVPGM? At least,
that way, presumably they could then invoke a program to issue:

RCLACTGRP ACTGRP(QILE)

periodically, as needed...

Or, is there a way to force the QZDASOINIT jobs to terminate, and
the replace them with a new instance, in the connection pool,
periodically?

Or any other recommendations?

There would be no point. The programs are coded to not cleanup for performance reasons, and reclaiming the activation group throws that all away. As Larry suggests the jobs should grow to meet the needs of the requesters, as coded, and eventually stop growing. When the job is ended, either by ENDJOB or reaching its maximum reuses, that will be the equivalent of the /reclaim/. Change the prestart configuration of the QZDASOINIT to have fewer reuses, and some of the jobs that have grown to accommodate serviced jobs [more quickly for having done so], will then tear down and be rebuilt; not to imply that is desirable.

Or as a friend always suggests that, probably, the wrong questions are being asked. In this case, I do not think the questions really should be about activation groups or reclaim thereof. I think if something concrete outside of the ACTGRP concept can be articulated, the real issues will become clearer.

For example consider a job that has seven pseudo-closed cursors sitting around after the first two uses. Is the concern that these are taking up resources? Is it not better that they remain to be used again, for performance, or is the system really suffering horribly for these remaining... and if so, what impact to what function?

Regards, Chuck