Recoup lost address processing

You can start processing lost addresses any time after the ZRECP RESUME command has run successfully, and before the next ZRECP START command is entered.

Lost address processing begins when you enter ZRECP LOST CREATE, which uses the lost address pseudo directories (#SONLOST) that were saved during recoup phase 3 to do the following:
  • Find all lost addresses
  • Record the lost addresses and their record IDs in the lost address database.
  • Calculate the total number of lost addresses for each record ID in each pool section and record them in the lost address database so that they can be displayed.

After ZRECP LOST CREATE processing ends, you can define a list of lost record IDs to be excluded from the return of lost addresses and remain protected when all other addresses are returned. If, after analysis, you determine that no IDs need to be excluded, you can bypass this step.

After you define any record IDs to be excluded, ZRECP LOST REBUILD processing rebuilds the lost address pseudo directories (#SONLOST) without the addresses of the lost records that are marked for exclusion. The rebuilt set of pseudo directories is used as input to pool directory update (PDU) processing to return the lost addresses.

Lost address database

The lost address database consists of the following parts:
  • A lost address summary file (LostAddressSummaryCount.fil), which resides in the file system directory specified by the PATH parameter on the ZRECP PROFILE command. If you do not specify a path name with the ZRECP PROFILE command, the file path defaults to /usr/IBM. This file contains the counts of lost addresses by record ID and pool section and indicators that define the lost addresses that will be included or excluded.
  • Details about the lost addresses, which are kept in #IBMLOST fixed file records. There is an entry for each lost address that includes the record ID and the record type and ordinal number for the file address.
    Note: If there are not enough #IBMLOST ordinals allocated, ZRECP LOST CREATE processing will record the details for lost addresses until all available ordinal numbers are full. When all #IBMLOST ordinals are full, processing will continue to identify the number of lost addresses by record ID and pool section in LostAddressSummaryCount.fil, but the details will not be recorded in #IBMLOST.

    If a record ID is excluded and details about one or more file addresses with that record ID were not recorded in the #IBMLOST record, all file addresses that do not have details in #IBMLOST will be excluded. Consider the following example: Record ID CD has 400 lost file addresses and 100 of these did not have details recorded in the #IBMLOST records, and record ID CA has 600 file addresses, 200 of which do not have details recorded in the #IBMLOST records. If record ID CD is excluded, the 300 detailed file addresses will be excluded, as well as all of the lost addresses without details, including the 100 file addresses with record ID CD and 200 file addresses with record ID CA.

    See z/TPF and z/TPFDF System Generation for more information about allocating #IBMLOST fixed file records.

Lost address PDU rollback

If an error occurs when you create and roll in the PDU to return the lost addresses, you can roll back the PDU to prevent database corruption. For more information, see Online PDU rollback support.