File Pair Verification

IBM® Connect:Direct® performs a verification procedure at initialization, as follows:

  • Within each file pair, IBM Connect:Direct verifies the appropriate sizing, relative to each other, of the entry-sequenced cluster and the key-sequenced cluster. If the file pair is not relatively sized, then IBM Connect:Direct issues a warning message and initialization continues.
  • If either of the files of a pair has data, IBM Connect:Direct attempts to verify that the two files are actually a statistics file pair. IBM Connect:Direct verifies that the key-sequenced file really does contain index information for the associated entry-sequenced file.

IBM Connect:Direct uses control information maintained in the key-sequenced file to perform the verification. The software keeps a control record in the KSDS which contains the data set name and the control interval size of the paired entry-sequenced file. If this information does not match, statistics initialization fails.

Changing the File Pair

Following are the implications of changing the file pair.

  • Changing the control interval size of the ESDS or renaming the clusters causes initialization to fail because the control record in the KSDS no longer matches the files. The two ways to resolve this statistics initialization failure are:
    • First, you can use the DGADBKEY utility to rebuild the key-sequenced cluster. This utility recreates the KSDS control record so that it matches the new names or control interval size. The records in the file pair remain accessible when the DTF is available again.
    • The second solution is to empty the files. The file pair is available for logging new records. However, the old records are no longer available. You may want to archive the files before emptying them.
  • Changing the size of a file pair is not a problem. The sizes of both files of a pair change together so that the relative sizes do not change.
    • If the files are made larger and the names remain the same, then copy the records from the old smaller entry and key-sequenced clusters to the new larger ones.
    • Use the same procedure to make a file pair smaller if all the existing records from both files will fit into the smaller space of the new files. If the existing records do not fit, then the new smaller file pair must be left empty initially, and the old records become unavailable.