On 14-Aug-2014 10:28 -0500, Steinmetz, Paul wrote:
Normally, only one joblog is created for a job.
Some jobs have multiple joblogs being created.
<<SNIP>>
In one instance 159 joblogs were created for a single batch job,
each about 80 pages, or 584k.
The final joblog is 5,164 pages, 32,792k.
FWiW, regarding the noted effect from *PRTWRAP setting for Job
Message Queue Full Action (JOBMSGQFL) having exceeded the Job Message
Queue Maximum Size (JOBMSGQMX):
<
http://www.ibm.com/support/knowledgecenter/api/content/ssw_ibm_i_71/apis/QMHCTLJL.htm>
_Control Job Log Output (QMHCTLJL) API_
"...
The Control Job Log Output (QMHCTLJL) API controls the production of a
job log when the related job ends or when the job message queue becomes
full and the print-wrap option is in effect for the job. The API can
only influence the job log for the job in which it is used. The API can
be used to control:
• The destination of the job log.
• The amount of message information written to the job log.
• The occurrence of messages in the job log.
To have an effect on the job log, the API must be called before the job
ends. It can be called in an initial program for the job.
This API does not produce a job log; rather, the API captures the
control information specified on the parameters and retains the
information until job log production. The API also prepares the output
files for job log production. When job log production occurs, such as at
end of job, the control information and the prepared output files direct
the job log to the output file.
If the API is not used, normal job log production occurs. That is, all
messages in the job message queue for the job are written to a spooled
file from which the job log can be printed. If the API is used, no
spooled file is produced. Instead, the messages are written into one or
two output files, depending upon the amount of information requested.
Once the API is used, the options selected remain in effect for the
current job until the API is called again. Each time the API is used,
the selected options can be changed. The options in effect when the job
ends or when the job message queue becomes full are the ones used to
produce the job log.
..."
Given the data is in a file, a summary query grouping on the MsgId
and counting the occurrences per message identifier is helpful to
determine which messages are pervasive and review if those might best be
eliminated; i.e. review of ~150 spooled joblogs is not nearly so simple
as a COUNT aggregate in an SQL GROUP BY query, optionally with a HAVING
clause and\or ORDER to limit the results to just the most common messages.
midrange.com
