Backup Options   «Prev  Next»

Recover Lost Control Files in Oracle 26ai - Exercise

Objective: Explain how to restore a binary control-file backup, recover the database and verify its return to service in Oracle AI Database 26ai.

This is a guided, written self-assessment, not an automatic grader or a command-execution tool. Study the diagrams, then explain your decisions in the response box. Use a disposable lab for any practical rehearsal.

Scenario and prerequisites

Both current control-file copies of the single-instance ORCL CDB have been lost. The instance is stopped and the storage fault has been repaired. The database is in ARCHIVELOG mode. All datafiles, required archived redo and online redo are intact; no structural changes occurred after the backup. The Oracle service and initialization configuration remain usable.

A verified binary backup exists at E:\Backup\ORCL\control.bkp. It was created before the failure using ALTER DATABASE BACKUP CONTROLFILE TO, not by copying a changing online control file. It is not an RMAN backup piece and not a text trace.

The existing CONTROL_FILES setting lists C:\OracleData\ORCL\control01.ctl and D:\OracleData\ORCL\control02.ctl. In this lab, C and D represent independent storage. Both directories exist and permit access by the Oracle service account. Run the illustrated shell commands in Windows Command Prompt using the correct Oracle home and an account authorized for local SYSDBA authentication. These are filesystem examples, not ASM or RAC instructions. Recovery applies to the CDB, not an individual PDB.

Guided recovery diagrams

Oracle 26ai: with the instance stopped, restore one binary control-file backup to both configured filesystem locations.
1. Confirm that no current copy survives. Restore the same binary backup to every configured location while the instance is stopped. If an intact current copy survives, replace the failed copy from it instead; that failure alone does not require backup-control-file recovery.
SET ORACLE_SID=ORCL
COPY "E:\Backup\ORCL\control.bkp" "C:\OracleData\ORCL\control01.ctl"
COPY "E:\Backup\ORCL\control.bkp" "D:\OracleData\ORCL\control02.ctl"
Oracle 26ai: connect with SQL Plus, mount, recover using backup controlfile, open resetlogs after recovery succeeds, then verify control files and CDB and PDB state.
2. Start SQL*Plus with sqlplus / as sysdba. Mount and perform recovery before opening. Apply requested redo; unresolved errors are a stop condition. Validate the open CDB and its PDBs, then take a fresh backup. The image is a guide, not evidence that recovery was executed.
STARTUP MOUNT;
RECOVER DATABASE USING BACKUP CONTROLFILE;

Only after recovery succeeds:

ALTER DATABASE OPEN RESETLOGS;
SELECT name FROM v$controlfile ORDER BY name;
SELECT open_mode, log_mode FROM v$database;
SHOW PDBS

Your recovery plan

Answer all six prompts. Compare your response with the model answer after submission; no numerical score is calculated.

  1. Why must you check for an intact current copy before restoring an older backup? How would that change the recovery path?
  2. State the shutdown and storage prerequisites. List the two copy destinations and explain why updating CONTROL_FILES alone is insufficient.
  3. Give the connection, mount and recovery commands for this scenario. Explain what to do if Oracle requests redo that is not in an archived log.
  4. Explain when RESETLOGS is appropriate here and what you would do if required redo were missing.
  5. Describe the control-file, database and PDB checks, plus the backup you would take after recovery.
  6. Explain why generating a trace and running CREATE CONTROLFILE are not required intermediate steps when this valid binary backup is available.

Review the control-file recovery lesson for the separate RMAN and trace-reconstruction paths.

Oracle 26ai references