On 4 Feb 2016 14:05:30 -0800, in bit.listserv.ibm-main you wrote: >At 13:16 -0700 on 02/04/2016, Sri h Kolusu wrote about Re: Count smf >type 30-5 records: > >>If you want DFSORT to extract the values from Raw SMF data then what >>fields correspond to Userid, Jobname(SMF30JBN ??) and Date from the SMF-30 >>Record mapping? Remember the raw smf have variable length segments where >>the offset and the length are stored in the record and you need >>dynamically access the information which would require an addition pass of >>data. > >Due to the variable length segment issue, just using Sort Cards to >map the file is not going to fly well. IMO, the simplest way of >extracting the needed fields is to just have someone who is capable >of writing simple Assembler code to just write a program to extract >the fields and output a simple file with them. This file can then be >sent through the sort which consolidates and counts them to output a >report whenever a field changes. It also can be done in COBOL. The program and data set usage program I wrote did it using reference modification.
Why IBM didn't go further with the JOBNAME, PROCNAME, STEPNAME fields and make the variable length up to 255 characters and allow multiple PROC names I don't know. How are accounting records done for tasks under UNIX systems services? Clark Morris > >I know that you are looking for a way to have sort handle all the >work, but doing the record selection/parsing is better delegated to >something that can locate the data. Back in the when I was still >coding, this type of task was one that I would toss off as a >Quick&Dirty project in an hour or two. > >---------------------------------------------------------------------- >For IBM-MAIN subscribe / signoff / archive access instructions, >send email to [email protected] with the message: INFO IBM-MAIN ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
