1、太太药业SAP备份及恢复流程SAP 备份及恢复流程Author: 陶朝晖Date: May.23, 2002Version: 1.0ContentPurpose 3Backup Strategy 3Backup Tool 3Backup Device 3Backup Contents 3Backup Scheme 4Tape Storage 5There are two ways to check status of backup 5Recovery Strategy 6Restore Tool 6Restore Device 6Restore Contents 6Recovery Schem
2、e 6Complete Recovery 6Incomplete Recovery 8Special cases 10Appendix 13I. Steps of SAP Archive 13II. Steps of SAP Online Backup 14III. Steps of SAP Offline Backup 16IV. Daily Recorder Report 18V . Contents in tape 19VI. Tape management 21VII. Structure Changes (Control Files ) 23VIII. ORA-01149 or OR
3、A-01113 for missing end backup 24PurposeThis document defines the backup policy to ensure regular, correct backup procedures are carried out and data integrity is kept.This document describes backup strategy. In the event of loss of data occurring, the system can therefore be restored as quickly and
4、 as fully as possible. This document also defines the recovery procedures to ensure system would be up as quickly and as fully as possible in case of system down. Backup StrategyThe recommended method is tape backup. Backup Tool Using the DBA Planning Calendar (transaction DB13) Using the SAP databa
5、se administration tool SAPDBA at operating system level Directly, at the operating system level ( brarchive/brbackup, tar/dump ) Using hp backup tools omni .Backup Device Tape driver: 24GB DLT Backup Contents Data Files Control files Profiles Online redo log files Offline redo log files Operating sy
6、stem filesTape Name Convention and Initialization Tape used for brbackup are always named as follows, B01, B02, B03, B04, B05, B06,Tape used for brarchive are always named as follows, A01, A02, A03, A04 A05, A06, Perform the tape initialization procedure as follows: Set the init.sap parameters volum
7、e_backup and volume_archive. In this example:volume_backup=(B01, B02, B03, B04)volume_archive=(A01, A02, A03, A04) To initialize the tape pools, use SAPDBA or BRBACKUP and BRARCHIVE with call option -i force (where -i = initialize and force = do not evaluate tape label). In this example, tape labels
8、 B01, B02, B03, B04 and A01, A02, A03, A04 are initialized. To overwrite the default settings defined in volume_backup or volume_archive, use BRBACKUP and BRARCHIVE with call option -v/-volume. Backup SchemeThe backup scheme should guarantee that production data be recovered within a day and develop
9、ment data be recovered within six days after a disaster.A systematic approach of maintaining backups on a daily, weekly, and monthly basis is strongly recommended: The first backup is always a Full backup. Daily: PRD: Online Backup ( night time ) + Redo log archive ( day time ) Other system: Redo lo
10、g archive ( day time ) Weekly: Offline Backup Monthly: Unix level BackupIf it is in following cases, backup under unix level must be carried out, R/3 release upgrade Kernel upgrade Oracle upgrade OS upgradeAll backups should be scheduled during off time, e.g. at the night of each day or at weekend.
11、The administrator should STICK TO the scheme. Occasionally, the administrator will have to schedule the scheme to run when no one will be around to insert the correct media into the tape drive, for example, on holidays. It is recommended that the administrator change the scheme so that he/she has a
12、current full backup of the set before the scheduled closing, then shut off backup for that day. When he returns, he can revert to the standard set-up for that set. The administrator should employ appropriate tapes to ensure they will hold all data needing backup. All tapes should be labeled with cer
13、tain SAP naming convention which will reflects Backup type and Date. Once the administrator have set up the rotation scheme, he need to make sure the right media is in the tape drive for each day of the week. The administrator should check backup logs after backup complete. Logs are very important i
14、n assuring data integrity. If the administrator needs to restore data in a disaster situation, hell have a reference from which to start. Tape StorageThree groups of tape media should be stored separately in appropriate place. A tape should not be rotated exceeding the guaranteed lifetime. For addit
15、ional security, the monthly tapes and Id floppy copies should be placed in a fireproof safe and stored off-site (Away from computer room).Verification and Failed BackupsAfter the backup has run, backed up files should be verified against the originals. To verify both the data to be backed up as well
16、 as the database backups, the following two ways check should be used, Perform a logical data check to verify the consistency of the Oracle database and void oracle blocks corrupted. For optimal performance, perform this check during periods of low system activity, such as weekends. The statement us
17、ed for checking is sapdba -check Perform a physical data check to verify the tapes used for a database backup. To check the physical correctness of the data transferred, read the tapes after a successful backup once a quarter.If a backup fails, the explanation needs to be found and the problem remed
18、ied before the next backup. In any case, the backup log should be examined. A problem may have arisen with the backup medium, the backup software or the system itself. The administrator should check every day that all backups have run successfully. There are two ways to check status of backup Using
19、the DBA Planning Calendar (transaction DB13) Using the Log info Overview ( transaction DB24 )Recovery StrategyIn this chapter, we only talk about recovery for SAP Data ( Database Data Files ). Restore Tool Using the SAP database administration tool SAPDBA at operating system level Directly, at the o
20、perating system level ( brarchive/brbackup, tar/dump )Restore DeviceTape driver: 40-80GB DLT Restore Contents Data Files Control files Profiles Online redo log files Offline redo log files Operating system filesRecovery SchemeThe backup scheme should guarantee that production data be recovered withi
21、n a day and development data be recovered within two days after a disaster.Complete Recovery General flow1. sqldba lmode=ySQLDBA SHUTDOWN NORMAL | IMMEDIATE | ABORTSQLDBA EXIT2. Make an OFFLINE backup of the whole DB, but at least of the current control file and those online redo-log files which wer
22、e not yet archived (see V$LOG) in case the recovery will have to be repeated!3. Restore only the defective data files from the last backup (whether ONLINE or OFFLINE backup). For the result, it is of no importance if all the data files are restored, but restoring and recovering single files is faste
23、r.( If there were structure changes since the last backup which affect the defective files, you must use the last backups AFTER the structure changes! Please refer to appendix )4. sqldba lmode=ySQLDBA CONNECT INTERNALSQLDBA STARTUP MOUNTSQLDBA ALTER DATABASE CREATE DATAFILE (with path) AS SIZE ;Crea
24、te all the files for which no backup exists, but only if all the redo-log-files since file generation exist and if there has been no CREATE CONTROLFILE since then (- ORA-01178). The AS option can be used if the new file should also be renamed, SIZE, if the file size was changed.SQLDBA SELECT NAME FR
25、OM V$DATAFILE WHERE STATUS IN (OFFLINE,SYSOFF);SQLDBA ALTER DATABASE DATAFILE ONLINE; (with path)Set all previously selected data files ONLINE!SQLDBA RECOVER DATABASE USING BACKUP CONTROLFILE;Provide redo-log-files (if needed) and confirm them with until Media recovery complete.is reported.USING BAC
26、KUP CONTROLFILE is necessary if a backup of the control file should be reloaded or ifCREATE CONTROLFILE was executed with option RESETLOGS.Generally, CREATE CONTROLFILE will be executed with NORESETLOGS (special case: e.g. change db_name, see ORACLE docu.).SQLDBA ALTER DATABASE OPEN RESETLOGS;RESETL
27、OGS is necessary after USING BACKUP CONTROLFILE.SQLDBA EXITThe DB is now completely recreated.If the DB was opened with RESETLOGS, you must shut it down at once and make a complete OFFLINE backup in order to be able to restore and recover in the future as well.Incomplete Recovery The following 3 opt
28、ions are available for an incomplete recovery: - up to a particular date and time (UNTIL TIME) - up to a particular SCN = system change number (UNTIL CHANGE) - up to a particular redo log file (UNTIL CANCEL)CAUTION: If you want to perform another recovery (i.e. import redo log files) after the resto
29、re (restore database backup), you may NOT OPEN the database until you have imported all the redo log files required; see number 4 further below.General Procedure:-1. sqldba lmode=y SQLDBA SHUTDOWN NORMAL | IMMEDIATE | ABORT SQLDBA EXIT2. Perform an OFFLINE backup of the entire database. At least sav
30、e the current control files and the not archived online redo log files (see V$LOG), in case the recovery needs to be repeated!3. Reload all the backed-up data files of the entire DB (restore). - Also reload undamaged files. All data files must have been saved BEFORE the recover-until time! - Excepti
31、on: Undamaged data files need NOT be reloaded, if despite choosing the option UNTIL the database is to be recovered to the end, that is to its current state. (e.g. When UNTIL only serves the purpose of enabling the database to be reopened with RESETLOGS after the recovery.) - For the recovery it is insignifican
copyright@ 2008-2022 冰豆网网站版权所有
经营许可证编号:鄂ICP备2022015515号-1