Veritas Storage Foundation™ 5.0.1 for Oracle RAC Installation, Configuration, and Administrator's Guide Extracts for the HP Serviceguard Storage Management Suite on HP-UX 11i v3

NOTE: Some database changes made after a Storage Checkpoint was taken may make it
impossible to perform an incomplete recovery of the databases after Storage Rollback of an online
or offline Storage Checkpoint using the current control files. For example, you cannot perform
incomplete recovery of the database to the point right before the control files have recorded the
addition or removal of datafiles. To provide recovery options, a backup copy of the control file
for the database is saved under the /etc/vx/vxdbed/$ORACLE_SID/checkpoint_dir/
CKPT_NAME directory immediately after a Storage Checkpoint is created. You can use this file
to assist with database recovery, if necessary. If possible, both ASCII and binary versions of the
control file will be left under the /etc/vx/vxdbed/$ORACLE_SID/checkpoint_dir/
CKPT_NAME directory. The binary version will be compressed to conserve space. Use extreme
caution when recovering your database using alternate control files.
Suppose a user deletes a table by mistake right after 4:00 p.m., and you want to recover the
database to a state just before the mistake. You created a Storage Checkpoint
(Checkpoint_903937870) while the database was running at 11:00 a.m., and you have ARCHIVELOG
mode enabled.
Guidelines for Oracle Recovery
For optimal Oracle recovery, follow these guidelines:
• Back up all control files before storage rollback in case the subsequent Oracle recovery is
not successful. Oracle recommends that you keep at least two copies of the control files for
each Oracle database and that you store the copies on different disks. It is also a good idea
to back up the control files before and after making structural changes to databases.
NOTE: The dbed_ckptcreate command automatically saves control file and log
information when you create a Storage Checkpoint.
• Make sure that the control files are not rolled back.
A control file is a small binary file that describes the structure of the database and must be
available to mount, open, and maintain the database. The control file stores all necessary
database file information, log file information, the name of the database, the timestamp of
database creation, and synchronization information, such as the Storage Checkpoint and
log-sequence information needed for recovery. Rolling back the control file will result in an
inconsistency between the physical database structure and the control file.
NOTE: If your intention is to roll back the database to recover from structural changes that
you do not want to maintain, you may want to use the backup control file that was created
by the dbed_ckptcreate command. The backup control file is located in the directory
$VXDBA_DBPATH/$ORACLE_SID/checkpoint_dir/CKPT_NAME.
• Make sure that all archived redo logs are available.
A database backup with online and archived logs is required for a complete database
recovery. Query V$ARCHIVED_LOG to list all the archived log information and
V$ARCHIVE_DEST to list the location of archive destinations.
To restore the necessary archived redo log files, you can query V$LOG_HISTORY to list all
the archived redo log history or query V$RECOVERY_LOG to list only the archived redo logs
needed for recovery. The required archived redo log files can be restored to the destination
specified in the LOG_ARCHIVE_DEST parameter or to an alternate location. If the archived
36 Using Storage Checkpoints and Storage Rollback