Point-in-time recovery
Use the PIT keyword to activate the point-in-time recovery option. See PIT keyword for more information.
The BMC Recovery Manager (RMGR), which is available with the BMC AMI Backup and Recovery for IMS product, can be helpful for point-in-time recoveries:
- You can use the RMGR Find Recovery Points function to scan the archived logs and look for a point in time when no updates were occurring and determine common windows of inactivity for all databases in a group.
- You can use the RMGR Log Analysis function to analyze the impact of the recovery time selection by displaying information about the update transaction activity in process during the specified point in time.
Considerations for point-in-time recoveries
If you specify a recovery time during which updates have taken place, it is probable that some activity will be incomplete and therefore will not be applied to maintain the database as logically correct.
All online activity, including BMPs, is applied through the commit point that is previous to or equal to the requested point in time. If the requested point in time is in the middle of a batch job, activity from the batch job is applied through the checkpoint that is previous to or equal to the requested point in time. If no checkpoint is taken in the batch job, no activity will be applied. If possible, select a recovery time during which update activity for the database is low.
Recovery log data set (RLDS) input is used for a PIT recovery.
Restrictions for point-in-time recoveries
If DBRC is defined for share control (SHARECTL) and you do not request an output image copy when you perform a special timestamp or point-in-time recovery, the Recovery utility automatically activates the IMAGE COPY NEEDED flag in DBRC and sets the job step completion code to 4.
The IMAGE COPY NEEDED flag means that you must create an image copy before the database can be updated.
If DBRC is defined for recovery control (RECOVCTL), you must create at least one output image copy during a special timestamp or point-in-time recovery.
As part of the recovery for IMS full-function databases, the Recovery utility writes both a RECOV record and a REORG record to the DBRC RECON data set. This ensures that future recoveries never attempt to use recovery inputs created prior to the special timestamp or point-in-time recovery. For DEDB databases, the Recovery utility writes only a RECOV record to the DBRC RECON data set.
- Rerun the special timestamp or point-in-time recovery specifying the exact timestamp as before.
- Perform a roll-forward recovery using only those logs created since the special timestamp or point-in-time recovery as input. You must specify DBRC(N).
- Create an image copy of the database. This image copy must be registered in the RECON data set.