Oracle DBA

In the high-stakes world of database reliability, disaster recovery must be validated before it’s ever needed. With Oracle 19c’s Fast-Start Failover (FSFO) feature, you gain powerful automation for failover scenarios under Data Guard. But how do you test failover logic without impacting your production setup? The answer lies in Observe-Only Mode, and that is the focus of this article.

What Is Observe-Only Mode?

Observe-Only Mode is a special operating mode for Fast-Start Failover where Oracle simulates the decision process for failover without actually executing it. In essence:

• No structural changes are made to your broker configuration.
• No switchover or failover is triggered.
• When a failover condition is met, the system logs what would have occurred, rather than enact it.

You get full visibility into how FSFO behaves — without risking your production environment.

Why Use Observe-Only Mode?

Observe-Only Mode offers several compelling benefits:

1. Risk-free testing
   You can validate how fast-start failover would behave in your environment—without affecting your databases or applications.
2. Visibility into failure conditions
   The Oracle Broker logs and observer logs record precisely when failover would have been triggered, and under what conditions.
3. Justification for FSFO adoption
   By observing failover decision points over time, you can build confidence (and justification) for enabling FSFO in “real” mode.

Key Behavioral Details in Observe-Only Mode

While in Observe-Only Mode, the system exhibits certain special behaviors:

• The primary database can enter UNSYNC or LAGGING states even if the observer or standby don’t acknowledge.
• The primary may be opened without acknowledgement from the observer or standby.
• Manual failover can be invoked even if pre-condition checks fail, including cases where:
  • Broker configuration is in UNSYNC or LAGGING state
  • The failover target is invalid
  • Reinstatement is in progress
  • A master observer switch is in progress
• You can still perform switchover or manual failover to a bystander database (i.e., not the FSFO target).

These behaviors allow you to “stress test” the tolerance of your system under various adverse conditions.

How to Enable and Use Observe-Only Mode

Here’s how you can enable Observe-Only Mode via DGMGRL:

DGMGRL> ENABLE FAST_START FAILOVER OBSERVE ONLY;
Enabled in Observe-Only Mode.

DGMGRL> SHOW CONFIGURATION;

Configuration – yourConfigName  
Protection Mode: MaxAvailability  
Members:  
  primaryDB – Primary database  
  standbyDB – Physical standby  
Fast-Start Failover: Enabled in Observe-Only Mode  
Configuration Status: SUCCESS  

If desired, you can disable this mode later with:

DGMGRL> DISABLE FAST_START FAILOVER;

Example Session

Below is an illustrative snippet of how this looks in practice:

DGMGRL> ENABLE FAST_START FAILOVER OBSERVE ONLY;
Enabled in Observe-Only Mode.

DGMGRL> SHOW CONFIGURATION;
Configuration – hari
Protection Mode: MaxAvailability
Members:
  delhi – Primary database
  chennai – (*) Physical standby database
Fast-Start Failover: Enabled in Observe-Only Mode
Configuration Status:
  SUCCESS (status updated 35 seconds ago)

DGMGRL> DISABLE FAST_START FAILOVER;
Disabled.

Best Practices & Suggestions

• Run Observe-Only Mode over a representative period (days or weeks), to capture normal fluctuations and anomalies.
• Review logs frequently (broker logs, observer logs) for conditions that would have triggered a failover.
• Use the insights to adjust DSIVEN, FSLO_TIME, or other related settings, so that your FSFO policy is aligned with your RTO / RPO goals.
• Only after you’ve built confidence via Observe-Only Mode, transition to a real failover mode in production.

By enabling Fast-Start Failover in Observe-Only Mode, you gain a dry run of how Oracle would behave under failure stress — with zero effect on operations. It’s a smart, pragmatic way to build trust in automation before going “live.”