When all nodes of the primary ApsaraDB RDS for PostgreSQL instance in a Global Active Database (GAD) instance group become unavailable, you can promote the secondary instance (disaster recovery instance) to the primary instance with one click to restore read/write access and resume business operations. This operation applies to disaster recovery drills and cross-region disaster recovery scenarios.
Prerequisites
Before you begin, make sure that the GAD instance group is in the Running state.
Usage notes
Potential data loss: Promoting a secondary instance to the primary instance may result in data loss. Proceed with caution.
Transient connection interruption: The connection to the instance group may be briefly interrupted during the promotion. Make sure that your application can automatically reconnect to the instance group.
Original primary instance is removed: After promotion, the original primary instance is removed from the instance group, and the one-way DTS synchronization task is stopped. The instance itself is not released — its data and the associated DTS synchronization account remain intact. You can continue to manage it on the Instances page.
After promotion, you can add the original primary instance back to the instance group as a new secondary instance. For more information, see Add or remove a secondary instance.
Promote the secondary instance
Go to the Global Active Instances page.
Click the ID of the instance group.
In the List of RDS Instances section at the bottom of the details page, click Upgrade to Primary Database in the Actions column for the secondary instance.
Read the notes in the dialog box and click OK.
After promotion
Once promotion is complete:
The secondary instance transitions from read-only to read/write state and becomes the new primary instance.
The original primary instance is removed from the instance group. It remains available on the Instances page and retains all existing data.
What's next
After the promotion completes, evaluate whether to update your application's connection configuration to use the endpoint of the new primary instance so that traffic is directed to the correct instance.