Choose an RMAN check from the question you need to answer: which pieces recovery needs, whether the selected material is readable, whether it is accessible, or which backups retention still requires. Each result leads to a different next action.
Oracle DBA Lesson 31D — Checking backups before recovery
The checks
| Check | Question and action |
|---|---|
RESTORE DATABASE PREVIEW SUMMARY; | Identify selected database backups and required redo from repository records. Locate their handles and arrange retrieval of offsite media. |
RESTORE DATABASE VALIDATE; | Read the selected database backups and check missing files and corruption while leaving data files in place. Review selected pieces and detailed errors. |
RESTORE ARCHIVELOG ALL VALIDATE; | Separately validate archived-log backup material in this lab. Match coverage to the actual recovery target. |
CROSSCHECK BACKUP; | Reconcile recorded availability through configured channels. Disk checks inspect headers. Tape checks query the media manager. Correct access before you interpret status. |
LIST EXPIRED BACKUP; | Review pieces still unavailable through those access paths. Check mounts, channels, and media configuration. Preserve the pieces while you investigate. |
SHOW RETENTION POLICY; / REPORT OBSOLETE; | Read the policy and review pieces it no longer requires. Check business retention, redo consumers, and independently tested copies before any cleanup decision. |
EXPIRED calls for an availability investigation. OBSOLETE calls for a retention review. A missing mount can change availability even when the file still exists. Correct access and crosscheck again. An accessible valid piece can return to AVAILABLE.
A recovery-window policy is evaluated from recovery dependencies. A file's age alone is not a retention decision. See RESTORE, validation, CROSSCHECK, and REPORT.
Practice
Use only the instructor's disposable 19c CDB and an authorized common backup operator connected to its root with SYSBACKUP or SYSDBA. Confirm the database and release update, root identity, backup storage, and channel or media access. Schedule validation's storage I/O. The database must be mounted or open for the restore validation shown here. Preserve all pieces and redo for the later isolated recovery drill.
- Capture
SHOW RETENTION POLICY, the preview, and the selected piece handles. - Run the database validation in the agreed I/O window. Preserve the complete log and named pieces. Validate the relevant archived-log material separately.
- With media access established, crosscheck backups and capture the expired inventory. Crosscheck updates repository status.
- Capture
REPORT OBSOLETEand explain the applicable policy. Do not delete. - Classify a supplied failure as missing inventory, unavailable media, corruption, or insufficient redo. Propose a concrete next investigation that does not change the database.
- Name the still-pending timed isolated restore, redo recovery, service check, and application check required for the agreed recovery-time goal.
You pass when you can choose the right check and next action, explain availability versus retention, preserve an EXPIRED piece during the investigation, and identify the restore and application test still required. Record the observed result, identity, time, errors, and interpretation. This practical does not change the retention policy and does not delete backups.
For the local OS-authenticated route, first verify the instructor-selected Oracle environment and OSBACKUPDBA membership. Start RMAN with a fresh evidence filename:
rman log=lab31d_checks_01.logAt the RMAN prompt, connect and capture the selection:
CONNECT TARGET "/ AS SYSBACKUP";
SHOW RETENTION POLICY;
RESTORE DATABASE PREVIEW SUMMARY;Check the connection's database identity against the selected lab and confirm the session is at the CDB root. An instructor-provided connection may be used instead. Keep its actual target identity in the log. Resolve every selected piece handle, required archived-log range, and offsite retrieval dependency. Preview selects from metadata. It does not read the backup blocks. See CONNECT.
In the agreed storage I/O window, run:
RESTORE DATABASE VALIDATE;
RESTORE ARCHIVELOG ALL VALIDATE;RESTORE ARCHIVELOG ALL VALIDATE checks the archived-log backup material RMAN selects in this lab. For an operational target, select and verify the required range. Do not treat this broad check as proof that the target range is covered. Record selected pieces, missing-media errors, and corruption details. Both commands use the configured devices and the current backup selection, so the channels must reach the intended media. They read backup material and leave the current data files in place.
With storage and media access established, inspect status and policy:
CROSSCHECK BACKUP;
LIST EXPIRED BACKUP;
REPORT OBSOLETE;
EXIT;Keep the log outside the database's failure domain. A temporarily unmounted backup disk can mark a piece EXPIRED. Restore access, rerun crosscheck, and check whether the piece returns to AVAILABLE. Missing headers or an inaccessible media manager need their own investigation. Do not create an outage or corrupt a piece for this exercise. Use instructor-supplied failure evidence. For OBSOLETE entries, record the actual policy and the separate business and redo-consumer review before you propose any cleanup.
For each check, record the target, time, selected handles or ranges, result, interpretation, and next action. Leave the timed restore, redo recovery, application acceptance, and measured recovery time until the isolated drill.
Recap
Preview the required material, validate readability, investigate availability, and review retention separately. A timed recovery drill is what supplies the application and recovery-time evidence.
No comments:
Post a Comment