Starting GTF

Mainframe Access provides a sample procedure, MFAGTF, for running GTF; its use is described in the section Adding the GTF Procedure to a System Procedure Library in the chapter Configuration. You may want to pre-allocate a permanent GTF trace data set on DASD that can be reused each time you start MFAGTF. If you do so, update the MFAGTF procedure to use this data set.

A PDS member is normally used to automate the specification of GTF options when GTF is started and the sample procedure uses parameters from another sample member, GTFCNTL. This sample is set up to start GTF for recording Mainframe Access' activity trace (USR '0E9' records) and VTAM's buffer trace (USR 'FEF' and 'FF1' records).

A sample system log for the startup of GTF is shown below. Remember to have Mainframe Access and other software components that are part of the tracing activity already running, connected and ready to go, if possible.

  S MFAGTF.MYGTF
  $HASP100 MFAGTF   ON STCINRDR
  IEF695I START MFAGTF WITH JOBNAME MFAGTF IS ASSIGNED TO USER RWITEK, GROUP CSIDVLP
  $HASP373 MFAGTF   STARTED
  IEF403I MFAGTF - STARTED - TIME=16.34.12
  AHL121I  TRACE OPTION INPUT INDICATED FROM MEMBER GTFCNTL  OF PDS
  <hlq>.CNTL
  TRACE=USRP
  USR=(FF1,FEF,0E9)
  END
  AHL103I  TRACE OPTIONS SELECTED --USR=(FEF,FF1,0E9)
  AHL906I THE OUTPUT BLOCK SIZE OF    27998 WILL BE USED FOR OUTPUT 997
          DATA SETS:
            <hlq>.TRACE
  AHL080I GTF STORAGE USED FOR GTF DATA: 998
          GTFBLOCK STORAGE       82K BYTES (BLOK=        40K)
          PRIVATE STORAGE      1038K BYTES
          SADMP HISTORY          54K BYTES (SADMP=       40K)
          SDUMP HISTORY          54K BYTES (SDUMP=       40K)
          ABEND DUMP DATA         0K BYTES (ABDUMP=       0K)
  AHL031I GTF INITIALIZATION COMPLETE
  

GTF is now up and running and ready to record type 'FEF' and 'FF1' records from VTAM. z/OS may also force some USR records into the trace data set for system components. GTF wraps trace data when writing trace records to a DASD data set. This means that you can leave GTF tracing on for a long time to catch an intermittent problem, then stop the trace soon after the problem occurs. However, if you do not stop the trace soon after the problem occurs, the trace data that you collected may be overwritten by trace data from other system components.