Why did I get a hardware maintenance or OS update notification for my Aurora PostgreSQL-Compatible DB instance?
Last updated: 2022-12-02
I received a maintenance notification altering me that one my Amazon Aurora DB clusters requires hardware maintenance or an OS update.
Amazon Relational Database Service (Amazon RDS) performs maintenance on your DB clusters to fix issues related to security and instance reliability. This includes Amazon Aurora DB instances such as Aurora MySQL-Compatible and Aurora PostgreSQL-Compatible instances. During the maintenance window, Amazon RDS applies updates related to hardware, operating systems, or DB engine minor versions. For more information, see Maintaining an Amazon Aurora DB cluster.
You can configure an Amazon CloudWatch Events rule to get notifications about maintenance actions on your Amazon Aurora DB instances. If you have already received a notification, then use the Amazon RDS console to view any pending maintenance actions for your DB instances.
Note: If you receive errors when running AWS Command Line Interface (AWS CLI) commands, make sure that you’re using the most recent version of the AWS CLI.
To get notifications about maintenance actions on your DB instance, see What do I need to know about the Amazon RDS maintenance window?
View maintenance actions for your Aurora DB instance
To view the maintenance actions for your Aurora DB instance, do the following:
- Open the Amazon RDS console.
- In the navigation pane, choose Databases.
- Choose the Aurora DB Cluster or DB instance that you want to view maintenance actions for.
- Choose the Maintenance & backups tab. You can view the list of pending maintenance actions under the Pending maintenance section.
You can also run the describe-pending-maintenance-actions AWS CLI command to list pending maintenance actions:
$ aws rds describe-pending-maintenance-actions --region example-region-name
Action notifications for hardware maintenance
You receive an email notification when hardware maintenance is scheduled on your DB instances. The notification includes the time frame in which the maintenance must be completed, the Availability Zones (AZ), and any resources that are affected.
You can take one of these actions after you receive a hardware maintenance notification:
- Apply the hardware maintenance immediately.
- If no action is taken, then the maintenance is automatically performed on the affected instance. The maintenance happens during the maintenance window and within the time frame mentioned in the notification.
Depending on how many instances that are in your cluster, you might experience downtime.
- Clusters with a single instance - During hardware maintenance, the cluster and the instance are unavailable for a few minutes until the hardware maintenance is complete.
- Clusters with at least two instances - If the affected instance is a writer instance, then the cluster and instances are unavailable for the time it takes the instance to failover. If the affected instance is a reader instance, then no failover is performed. In this case, the reader instance is unavailable for a few minutes until the maintenance is complete.
Note: Hardware maintenance actions can't be deferred.
Action notifications for OS maintenance
The Amazon RDS upgrade to the operation system to improve performance takes about ten minutes to complete. This doesn't change your DB engine version or the instance class of your Amazon Aurora DB instance. Amazon RDS applies its mandatory updates automatically, on a schedule. You might receive notifications about scheduled OS maintenance such as this. For optional updates, Amazon RDS doesn't apply these updates automatically. But it's a best practice to keep your Amazon Aurora DB clusters up to date.
Note: It's a best practice to apply updates to a DB cluster before you apply them on the reader DB instances. Then, apply updates on the writer DB instance in order to avoid the excess downtime. If the OS update is applied on the reader, then the reader is unavailable until the update is complete. When the writer is updated, a failover is performed. The DB instance is unavailable for the time that it takes the instance to failover (usually about 60 seconds).