Stopping and starting an Amazon Aurora DB cluster

Stopping and starting Aurora DB clusters helps you manage costs for development and test environments. You can temporarily stop all the DB instances in your cluster, instead of setting up and tearing down all the DB instances each time that you use the cluster.

Overview of stopping and starting an Aurora DB cluster

During periods where you don't need an Aurora DB cluster, you can stop all instances in that cluster at once. You can start the cluster again anytime you need to use it. Starting and stopping simplifies the setup and teardown processes for clusters used for development, testing, or similar activities that don't require continuous availability. You can perform all the Amazon Web Services Management Console procedures involved with only a single action, regardless of how many instances are in the cluster.

While your DB cluster is stopped, you're charged only for cluster storage, manual snapshots, and automated backup storage within your specified retention window. You aren't charged for any DB instance hours.

To minimize charges for a lightly loaded Aurora cluster, you can stop the cluster instead of deleting all of its Aurora Replicas. For clusters with more than one or two instances, frequently deleting and recreating the DB instances is only practical using the Amazon CLI or Amazon RDS API. Such a sequence of operations can also be difficult to perform in the right order, for example to delete all Aurora Replicas before deleting the primary instance to avoid activating the failover mechanism.

Don't use starting and stopping if you need to keep your DB cluster running but it has more capacity than you need. If your cluster is too costly or not very busy, delete one or more DB instances or change all your DB instances to a small instance class. You can't stop an individual Aurora DB instance.

The time to stop your DB cluster varies depending on factors such as DB instance classes, network state, DB engine type, and database state. The process can take several minutes. The Amazon RDS service performs the following actions:

The time to restart your DB cluster varies depending on factors such as database size, DB instance classes, network state, DB engine type, and the database state when the cluster was shut down. The startup process can take minutes to hours, but usually takes several minutes. We recommend that you consider the variability in startup time when creating your availability plan.

To start the DB cluster, the service performs actions such as the following:

Limitations for stopping and starting Aurora DB clusters

Some Aurora clusters can't be stopped and started:

Stopping an Aurora DB cluster

To use an Aurora DB cluster or perform administration, you always begin with a running Aurora DB cluster, then stop the cluster, and then start the cluster again. While your cluster is stopped, you are charged for cluster storage, manual snapshots, and automated backup storage within your specified retention window, but not for DB instance hours.

The stop operation stops the Aurora Replica instances first, then the primary instance, to avoid activating the failover mechanism.

aws rds stop-db-cluster --db-cluster-identifier mydbcluster

Possible operations while an Aurora DB cluster is stopped

While an Aurora cluster is stopped, you can do a point-in-time restore to any point within your specified automated backup retention window. For details about doing a point-in-time restore, see Restoring data.

You can't modify the configuration of an Aurora DB cluster, or any of its DB instances, while the cluster is stopped. You also can't add or remove DB instances from the cluster, or delete the cluster if it still has any associated DB instances. You must start the cluster before performing any such administrative actions.

Stopping a DB cluster removes pending actions, except for the DB cluster parameter group or for the DB parameter groups of the DB cluster instances.

Aurora applies any scheduled maintenance to your stopped cluster after it's started again. Remember that after seven days, Aurora automatically starts any stopped clusters so that they don't fall too far behind in their maintenance status.

Aurora also doesn't perform any automated backups, because the underlying data can't change while the cluster is stopped. Aurora doesn't extend the backup retention period while the cluster is stopped.

Starting an Aurora DB cluster

You always start an Aurora DB cluster beginning with an Aurora cluster that is already in the stopped state. When you start the cluster, all its DB instances become available again. The cluster keeps its configuration settings such as endpoints, parameter groups, and VPC security groups.

Starting your DB cluster usually takes several minutes.

aws rds start-db-cluster --db-cluster-identifier mydbcluster