A DBA needs to be able to identify the types of failures that
can occur in an Oracle Database environment. Using log and trace
filesk, a DBA can diagnose backup and recovery problems and
effectively find a solution.
Type of failures
Statement failure (bad data type, insufficient space)
action=>Increase quota limits
ALTER USER USER_NAME
DEFAULT TABLESPACE USER_DATA
TEMPORARY TABLESPACE TEMPORARY_DATA
QUOTA 2M ON USER_DATA
QUOTA UNLIMITED ON USER_INDEX
QUOTA 10M ON TEMPORARY_DATA
action=>Add datafiles to a tablespace or resize datafiles
Syntax for Creating a tablespace with and associating it with a datafile
create tablespace (tablespace_name)
DATAFILE 'C:\ORANT\DATABASE\usr1.dbf' SIZE 200M
DEFAULT STORAGE (INITIAL 500K NEXT 500K
MAXEXTENTS UNLIMITED PCTINCREASE 0)
PERMANENT
ONLINE
Syntax for Altering the size of a datafile actively being used by a tablespace
ALTER DATABASE
DATAFILE 'C:\ORANT\DATABASE\usr1.dbf' RESIZE 70M
Syntax for Moving a data file
alter database datafile
'c:\data\datafile1.dbf' offline
alter tablespace myTablespace offline
Alter database rename file
'c:\data\datafile1.dbf' to
'd:\data\datafile2.dbf';
alter database datafile
'c:\data\datafile1.dbf' online
select * from v$datafile;
(determine the tablespace to be taken offline)
alter tablespace myTablespace online
Take the tablespace for the datafile offline before renaming the datafile.
Insufficient privileges
action=>Grant roles or privileges
Example: Object Privileges to a role
grant execute on myFunction to a_user_role;
grant select,update,delete,
insert on myTable to a_user_roles;
User Process Failure
- User performed an abnormal disconnect
- User's session was abnormally terminated
- User's program raised an address exception
PMON detects abnormal termination
PMON rolls back the transaction and releases resources and locks.
User Error
- User drops a table
- User damages data by modification
action=>Point-in-time data recovery
Method 1 - use the imp utility to restore a table
Method 2
syntax of the recover [automatic] database
- option =
- until time 'YYYY-MM-DD:HH:MI:SS'
until cancel
until scn
using backup controlfile
Steps
Datafiles from from the last backup should be archived
Restore all datafiles
Do not restore control files, redo logs, passwords, or parameter files
Mount the database and recover the datafiles to a time before the failure occurs
startup MOUNT pfile='initxx.ora';
recover database until time '1999-1-1:10:30:00';
Open the database with resetlogs
alter database open resetlogs
Perform a closed database backup
shutdown normal
Backup archived logs from the system to prevent mixing archives from different database incarnations.
Instance failure
Start the Instance
startup pfile='initxx.ora';
Verify the instance is running
select * from V$SGA
SMON performs roll forward process by applying changes recorded in the online redo log files from the last checkpoint.
Notify users the Instance is running
Media failure
Disk Drive Failure
Data File has been erased
Alert Logs
- Records informational and error messages
- All Instance startups and shutdowns are recorded in the log
- Every Create, Alter, or Drop operation on a rollback segment, tablespace, or database is record in the log
The BACKGROUND_DUMP_DEST initialization parameter tells where the alert log and trace files are to be put.
Setting the LOG_BLOCK_CHECKSUM to be TRUE will enable redo log block checking.
Recovery Views
after running
ALTER DATABASE RECOVERY
USE THE FOLLOWING four views to determine status
V$RECOVER_LOG,
V$RECOVER_FILE,
V$RECOVER_FILE_STATUS,
$V$RECOVER_STATUS
DB Verify
Used to insure that a datafile is valid before a restore