Previous Topic: Split File Processing

Next Topic: Split File Time Processing

Split File Index Processing

The split file index file has the following characteristics:

RECFM=FB or F
LRECL=192
DSORG=PS
BLKSIZE=192 or a multiple of 192

To have CA SMF Director use a split file index, the file must be created and cataloged prior to the first dump process that accesses the index. CA SMF Director will not create the index file while running. The index file is dynamically allocated with DISP=OLD when being processed.

The index is a character format file, and can be read by column or by using a space character as a delimiter. The layout is available in member $SPINDEX in CASFMAC.

There are four possible states for a split file indicated in the index. In addition, diagnostic records that do not contain information about a split file may also appear in the index. These diagnostic records provide information that can be used on an EXTRACT process to assist in recreating a split file that has failed.

The split file states are as follows:

Complete (C)

The split file has all of the records it will contain. No further records will be written to it. The file is ready for post processing. Files are marked Complete at the end of a dump process when there is no TIME operand on the SPLIT statement, or when a boundary is crossed as indicated by a TIME or TIMECLOSE operand.

Empty (E)

This split file has no records in it, and will not have any records added to it. This is a special form of the Complete state that can be checked on by post-processors so that the file can be skipped.

Incomplete (I)

This state indicates a split file was created when there was a TIME operand on the SPLIT statement. The file was the last file allocated during the dump process for the SPLIT statement, and more records might be added to this file during the next dump process.

Abnormally Terminated (X)

This state indicates that an abnormal termination was encountered when generating a split file. The statistics in the entry indicate the number of records written to the split file and the time range of the records when the error occurred. The ABEND code is also reported.

When an abnormal termination record appears in the split file index, it is followed by a diagnostic record. A diagnostic record is indicated with a state of D. The information in the abnormal termination record and diagnostic record can be used to create an EXTRACT statement to recreate the failed split file.

At termination time, if the INDEXDUP operand is indicated on the SPLIT statement, a copy of the index will be written to the ddname indicated in the INDEXDUP operand.

Note: If multiple SPLIT statements are sharing a single split file index, the INDEXDUP operand only needs to be on the last SPLIT statement that indicates that index.