RDS - Known Issue - My RDS DB instance is in the failed state

Solution:

The failed state for RDS DB instances usually indicates that the hardware associated with your RDS DB instance has failed, and that AWS wasn’t able to recover it.

If you previously configured an automatic or manual backup for your RDS DB instance, you can restore your instance from that backup, or you could restore your instance using point-in-time recovery (PITR).

Note: PITR is only available for RDS DB instances that have automatic backups configured (the backup retention period is not 0). To retain the same endpoint name for your RDS DB instance, perform a PITR of your instance using a different name before deleting the instance in the failed state.

After recovering your instance by using your preferred backup solution, you can delete the failed instance.

After the failed DB instance has been deleted, you can optionally rename the newly restored DB instance using the old DB instance's endpoint name, which will eliminate the need to change the configuration of your application to connect to the new DB instance.

To help mitigate failures, consider running your RDS instances in a Multi-AZ deployment, and keep RDS best practices in mind when designing your architecture.