How backup monitoring works: backup age and failed backups

Applies to: Opslogix Oracle Management Pack (all versions; the Failed Backups monitor from V26.11.4544.0)
Last updated: November 2026 · Latest version: 26.11.4544.0

See also: Upgrade guide for 26.11.4544.0 · Version support · Release notes

The Oracle Management Pack discovers one backup object per Oracle instance, named "Backup state for <database name>". The backup monitors run on this object, on primary and standby databases. You find them in the backup state and alert views of the Oracle folder in the Monitoring pane.

Only RMAN backups

All backup monitors read V$RMAN_STATUS, the history of RMAN jobs that Oracle keeps in the control file. They see every backup made with RMAN, also when backup software starts RMAN through a media manager. Backups made in another way, for example storage snapshots or copies of the data files, are not in V$RMAN_STATUS, so the monitors don't see them.

The monitoring account needs SELECT on V$RMAN_STATUS. It is in the grant list of KB: How to create the monitoring account for the Oracle monitoring.

Backup age monitors

Seven monitors check the age of the last successful backup of one RMAN backup type each:

They run every hour. A monitor turns Warning when the last successful backup of its type is more than 25 hours old, and Critical when it is 99 hours old or more. Change these values (in hours) with an override on HealthyThreshold and CriticalThreshold, for example for a weekly full backup.

No RMAN backup: when V$RMAN_STATUS has no successful backup of a type, for example because RMAN never made one, the monitor of that type stays Healthy. This is by design: organizations that use another backup method, or that do not back up some databases, would otherwise get an alert for every monitored database. As soon as RMAN has made a backup of that type, the monitor checks its age. If a database has old RMAN backups but is now backed up in another way, disable the backup monitors for that database with an override.

Failed Backups monitor

The age monitors only notice a failed backup when the last successful backup becomes too old, which can be a day later. From V26.11.4544.0 the Failed Backups monitor reports a failure at its next check:

• It runs every 15 minutes and looks at the backup runs that ended in the last 48 hours (override LookbackHours).
• For every backup type, for example DB FULL, DB INCR or ARCHIVELOG, the latest finished run decides. Runs that are still running are skipped, so a retry in progress does not clear a failure.
• Critical when the latest run of a type is FAILED or COMPLETED WITH ERRORS. The alert "Oracle RMAN backup failed" lists the failed types with their status and start time.
• The alert closes by itself after a later run of each failed type succeeded.

Not reported: a failed run that is followed by a successful run of the same type between two checks, and a run that fails before the backup command starts, for example when RMAN cannot connect or allocate a channel. The backup age monitors report the second case.

When a backup failed

Find the failed runs, then read the RMAN output of that session. V$RMAN_OUTPUT keeps the output since the instance started.

SELECT session_recid, object_type, status, start_time FROM v$rman_status
WHERE operation = 'BACKUP' AND status IN ('FAILED', 'COMPLETED WITH ERRORS')
ORDER BY start_time DESC;

SELECT output FROM v$rman_output WHERE session_recid = <session_recid> ORDER BY recid;

Common causes:

• Space: the backup destination or the fast recovery area is full. Free space on the destination, or delete obsolete backups with RMAN> REPORT OBSOLETE; and RMAN> DELETE OBSOLETE;
• Destination: the backup destination is not available or not writable, for example an unmounted file system or a media manager error.
• Files: a data file or archived redo log is missing or corrupt, for example archived logs that were deleted outside RMAN.

After you fixed the cause, run the backup job again. The alert closes at the next check after the run succeeded.

Data Guard standby databases

The backup monitors also run on standby databases, including mounted ones, because V$RMAN_STATUS is available there. A standby only shows the backups made on that standby. If you back up only the primary, or only the standby, disable the backup monitors on the other database with an override.

Overrides

See How to change thresholds and intervals with overrides.