Why did my Amazon RDS for Oracle DB instance restart, recover, or fail over?

3 minute read

My Amazon Relational Database Service (Amazon RDS) for Oracle DB instance automatically restarted, recovered, or failed over.

Short description

Amazon RDS for Oracle DB instances can automatically restart, recover, or fail over for the following reasons:

There are underlying infrastructure issues in the DB instance. For more information, see Failing over a Multi-AZ instance for Amazon RDS.
An increase in workload caused a performance bottleneck and resource contention. To monitor the DB load, use tools such as, Amazon CloudWatch Metrics, Enhanced Monitoring, and Performance Insights.
The DB instance engine version was upgraded, or mandatory OS updates were made during the OS maintenance window. For more information, see Oracle upgrades in a Multi-AZ deployment.
You used the Reboot or Reboot with failover options to manually reboot the DB instance.
You added specific options to the DB instance option group that caused downtime. A brief outage occurs when you add or remove options such as, Oracle JVM, time zone option, and time zone file auto upgrade. For more information, see Adding options to Oracle DB instances.
You modified the DB instance to apply specific changes that caused downtime. To review the settings and changes that cause downtime, see Settings for DB instances.
The DB instance automatically restarted after 7 days because the instance was in the Stopped status.

To minimize issues with automatic restarts, recoveries, or failovers on your DB instance, use the following best practices:

Review the complete log of Amazon RDS events from the past 24 hours. To keep event data for an extended time period, it's a best practice to forward Amazon RDS events to CloudWatch Events. Use the logs to help you identify the underlying reason for the instance restart.
Activate Multi-AZ deployment on your DB instance to reduce downtime during an outage.
Create an event subscription to Amazon RDS event notifications to notify you when your DB instance restarts.
Configure CloudWatch alarms for key CloudWatch metrics that show instance availability. Use metrics such as, CPUUtilization, FreeableMemory, ReadIOPS, WriteIOPS, SwapUsage, and FreeStorageSpace.
Make sure that your queries are correctly tuned to optimize database performance. Otherwise, you might experience performance issues and extended wait times. Use the Performance Insights dashboard to analyze Oracle Execution plans and wait events.

Topics