Considerations for the FRRECOV command
The following topics are related to your use of the FRRECOV command:
- Before processing an FRRECOV COPYPOOL command: DFSMSdss
restores a volume only if the volume is not allocated to another job.
If the volume contains a catalog, the catalog address space in a DFSMSdfp
environment may have the catalog (and the volume) allocated. If the
copy pool catalog names are specified in the SMS copy pool definition,
DFSMShsm attempts to unallocate the catalogs prior to recovery. If
the catalog names are not specified in the copy pool definition or
the unallocation fails, follow these steps:
- Issue the F CATALOG,OPEN(volser) command to determine which catalogs on the target volume are allocated.
- For each catalog that is open, issue the F CATALOG,UNALLOCATE (catalogname) to cause the catalog address space to close the catalog and deallocate it. When the last catalog on the volume is deallocated, the target volume is also deallocated by the catalog address space.
- Issue a D U,,,addr,1 command to ensure that the volume is not allocated by another user.
- If the volume is not marked as allocated, issue the FRRECOV command to DFSMShsm.
- Processing an FRRECOV command: When SnapShot
is used as the fast replication utility, a recovery can be performed
immediately after the DFSMShsm FRBACKUP command completes.
When FlashCopy® is used as the fast replication utility, a recovery from a fast replication backup cannot be performed if any of the volumes are in an existing FlashCopy relationship unless fast reverse restore is to be used. If fast reverse restore is not intended, do not attempt to recover the copy pool until all of the background copies have completed from an invocation of the FRBACKUP command. You can use the QUERY COPYPOOL command to determine if any of the source volumes for a copy pool are in a FlashCopy relationship.
When a copy pool is defined without fast reverse restore capability, you can use the FRRECOV command with the VERIFY keyword to prevent the initiation of a recovery of a copy pool for which all of the background copies from the backup have not yet completed. When you specify VERIFY(Y), before initiating any volume recoveries, DFSMShsm first determines if any of the volumes are in an existing FlashCopy relationship. If so, the recover fails. If you need an immediate recovery, you have two options:
- Wait until all of the background copies have completed. Then the latest backup version can be used for the recovery.
- For FlashCopy version
1, withdraw the outstanding FlashCopy relationships using the FRBACKUP COPYPOOL(cpname) WITHDRAW command. Doing this invalidates the latest version
of the backup and enables an immediate recovery from the prior version.
This process only works for FlashCopy version 2 when the generation 1 background copies are complete.
Attention: Use the LIST COPYPOOL command or the ARCXTRCT macro to ensure that you do not invalidate the only valid fast replication backup version.
When a copy pool is defined with fast reverse restore capability, the VERIFY(Y) keyword allows DFSMShsm to determine whether fast reverse restore or regular FlashCopy can be used. For more information, see Recovering a copy pool using fast reverse restore.
- TSO usage considerations: When you use commands
that specify WAIT or NOWAIT, the results sent back to the user depend
on whether the WAIT parameter is specified and the success or failure
of one or more volumes:
- If WAIT is specified and all volumes process successfully, a return code of zero is returned.
- If WAIT is specified and one or more volumes fail, a nonzero return code and associated messages are returned.
- If WAIT is not specified, control is returned to the system after the command has been accepted and verified. All messages continue to be issued, but you are not notified when the command completes.
- Trying a copy pool recovery again: During
recovery of a copy pool, one or more volumes might fail to be recovered.
In order to complete the recovery, you must identify and correct the
cause of the failures. After correcting the cause of the failures,
your options for completing recovery depend on the source of the recovery
(DASD or tape) and if fast reverse restore was used.
- When the copy pool was recovered from DASD without using fast
reverse restore (that is, the background copy completed before the
recovery began), you can try to recover the individual volumes that
failed, or try to recover the entire copy pool again.
- If a small number of volumes failed recovery, issue a FRRECOV COPYPOOL(cpname) TOVOLUME(volser) command for each volume to try recovery of that volume again.
- If FlashCopy version 2 or SnapShot is your fast replication utility, you can immediately try to recover the entire copy pool again.
- If FlashCopy version 1 is your fast replication utility, you must wait for the FlashCopy relationships that were successfully established during the prior recovery to complete before you can try to recover the copy pool again.
- When the copy pool was recovered from DASD using fast reverse restore or the recovery was from tape, the recovery can be resumed. This is possible because the recovery status of each volume in the copy pool is tracked in the CDS records. The LIST COPYPOOL(cpname) command can be used to determine if a copy pool version previously failed recovery. If the recovery of a copy pool version failed, the LIST COPYPOOL output will indicate "RECOVERYINCOMPLETE=Y" and, if fast reverse restore was used, the fast replication state will be FCFRRINCOMPLETE. To resume recovery of the copy pool, issue a FRRECOV COPYPOOL(cpname) command specifying the same token, version, date, or generation as the failed command. The volumes that failed during the prior recovery are determined and a recovery is attempted for only those volumes that failed.
Note:- If the initial copy pool recovery was from a DASD version using fast reverse restore (the FASTREPLICATIONSTATE is FCFRRINCOMPLETE) and the recovery is resumed from a dump version, the default is to resume recovery of the remaining volumes. You can specify RESUME(NO) to perform a full recovery from tape. The copy pool DASD backup version will be invalidated when the copy pool recovery is completed successfully.
- If the initial copy pool recovery was from a dump tape version and you are now trying to recover from the DASD version, a full recovery is performed.
- The VTOC and the volume serial on the target volume may have changed as a result of the command. Before the volume can be accessed on any remote system, the UCB must be refreshed. The refresh occurs automatically if the volume is online and the device manager REFUCB function is enabled. You enable the REFUCB function through PARMLIB member DEVSUPxx or the MODIFY DEVMAN command. For more information, refer to the description of the REFUCB keyword in z/OS MVS Initialization and Tuning Reference or z/OS MVS System Commands.
- When the copy pool was recovered from DASD without using fast
reverse restore (that is, the background copy completed before the
recovery began), you can try to recover the individual volumes that
failed, or try to recover the entire copy pool again.
Related reading
- For more information about FRRECOV and the parameters for recovering fast replication backups, see FRRECOV command: Requesting a fast replication recovery.
- For more information about recovering a copy pool using fast reverse restore, see Recovering a copy pool using fast reverse restore.