4. Operation › 4.10 Integration with CA SMF Director › 4.10.5 CA SMF Director Usage Notes

4.10.5 CA SMF Director Usage Notes

 Based on the definition, CA MICS components read the CA SMF
 Director duplicate split indices and selects ALL previously
 unseen CA SMF split data sets (files) as input into the DAILY
 update process.
 
 The success of this processing is based on the following:
 
 o  CA SMF Director is being executed on a continual basis.
 
 o  When interval data is being managed by a TIME-based split
    file, you might need to adjust the times that an SMF
    switch is issued to ensure that the interval data is
    available in a timely fashion.  When SMF Director writes
    TIME-based split files, it will not indicate that a file
    is ready for processing until a record for that split file
    appears in the next interval that the TIME statement
    indicates.  In order to ensure that the data is
    immediately available, make sure that the scheduled time
    for SMF switch operations are set to the times defined in
    the TIME operand plus the value of the INTVAL in the
    active SMFPRMxx member, plus one minute.  By doing this,
    there should be data that falls into the next interval.
 
    For example, suppose a SPLIT statement has the following
    TIME operand: TIME(0000,0800,1600), which indicates that
    there should be three iterations of the split file per
    day:
 
    - one with data from midnight to 8 AM
    - one with data from 8 AM to 4 PM
    - one with data from 4 PM until the following midnight
 
    Also suppose that the active SMFPRMxx member has
    INTVAL(15) defined.  This would cause all interval data to
    be generated every 15 minutes.
 
    Using this example, you would set automated SMF Switch
    commands (I SMF) to be issued at 00:16, 08:16 and 16:16
    every day.  This way, once the dump is completed from the
    switch, the data from the previous portion of the day
    would be immediately available.
 
 o  CA SMF split data sets are cataloged.
 
 o  If the split data sets reside on tape, your installation
    allows dynamic allocation of tape data sets.
 
 o  You are recording different SMF data types to unique CA
    SMF Director duplicate split indices, or plan to use
    DAYSMF.
 
    For example, if you are writing SMF and CICS data to
    unique split data sets, but recording their creation in
    the same CA SMF duplicate split index, CA MICS will read
    all the eligible files in the index.  That is, those data
    sets that satisfy the date and time range criterion of not
    previously processed; all records are read regardless of
    the component step unless it is DAYSMF, which will further
    split the data accordingly.
 
 Important:  Complete the worksheet in section 4.10.4.1 and
 review the implementation scenarios in section 4.10.3 to find
 the implementation that best describes the way CA MICS
 operates at your installation.  Tailor CA SMF Director
 accordingly.
 
 If the split data set creation fails for any reason, the CA
 SMF Director DUMP job will complete with a return code 4,
 record diagnostic messages to SYSPRINT, and update the index
 and duplicate split index with the data set name, date and
 time range, etc. of the SMF data processed including a split
 file state code of 'X', meaning abnormally terminated.
 
 If CA MICS encounters a never-before-seen split file state
 code 'X', it checks further to see if the data for that same
 date and time range has been recovered.  If so, processing
 continues.  If not, CA MICS writes an error message to the
 MICSLOG indicating that CA SMF Director failed to create the
 requested split data set, and terminates the job.  For
 details on recovering a failed SMF split data set, see
 section 4.10.4.3.
 
 An occasion may arise such that there is a need to bypass the
 processing of a split file.  For example, the corresponding
 split data set may contain a bad SMF record that causes the
 CA MICS DAILY job to fail, and you wish to ignore this file
 so as to continue processing other input data sets.  You can
 mark an index record in the duplicate split index as unusable
 (to be bypassed by MICS-SMFD interface) by changing its file
 state code located in column 55 to 'E', meaning empty.  To
 allow this file to be processed later, change the state code
 to 'C'.