Database Reference
In-Depth Information
it will continue to search for newer versions of the corrupted blocks. If a newer version is
available, the restore and recovery can continue. If there are no uncorrupted newer versions
of the block, the operation will fail.
To perform BMR, the database must be open or mounted and in archive log mode
and must have a current, usable control file. The database must not be a standby data-
base. You must use level 0 or full backups for the restore. All of the required archived
redo logs must be available for the recovery process.
If you have enabled Flashback Database and logging, then RMAN will search the
flashback logs for uncorrupted versions of the required blocks.
The steps to recover blocks using BMR are fairly simple. From SQL*Plus, determine
which blocks need recovery by viewing the alert log or querying the
V$DATABASE_BLOCK_
CORRUPTION
view:
SQL> SELECT FILE#, BLOCK#, BLOCKS, CORRUPTION_TYPE “TYPE”
FROM V$DATABASE_BLOCK_CORRUPTION;
FILE# BLOCK# BLOCKS CORRUPTION_CHANGE# TYPE
---------- ---------- ---------- ------------------ ---------
1201 1968 2 PHYSICAL
Now that you have the blocks required for recovery, connect to the target database with
RMAN and begin the recovery:
RMAN> RECOVER DATAFILE 1201 BLOCK 1968;
RMAN also allows you to recover all corrupt blocks in a database using BMR. Query
the
V$DATABASE_BLOCK_CORRUPTION
view to measure the extent of the damage; then launch
RMAN to perform the recovery:
RMAN> RECOVER CORRUPTION LIST;
When a block is repaired, it is removed from the
V$DATABASE_BLOCK_CORRUPTION
view.
Block media recovery will fail if there is physical corruption in the redo logs
that results in a checksum failure.
Managing Database Performance
In the following sections, you will learn how to monitor operations and performance in a
container database (CDB) and pluggable databases (PDBs) and manage allocation of resources
between PDBs and within a PDB. To accomplish these tasks in the Oracle 12
c
multitenant
Search WWH ::
Custom Search