Stop RDS more than 7 days

Scenario

If RDS needs to be shut down for an extended period without use, it will automatically start up every 7 days, after which it needs to be manually shut down. Therefore, using Lambda, we can periodically start specific RDS instances with certain tags for 30 minutes every 7 days and then shut them down again, achieving this purpose.

Implementation Method

Create an IAM Role to provide permissions for starting and stopping RDS
Set tags for RDS instances that need to be shut down long-term, and adjust all RDS maintenance times to be consistent
Create a Lambda Function and set up periodic start/stop triggers
Test whether it can start and stop according to the settings

Execution Steps

Create IAM Role

Create Policy

Click here to open the IAM Policy settings screen, then click Create Policy for subsequent Role use

Create IAM Policy

Click on JSON, paste the following Policy, then click Next

1
{
2
    "Version": "2012-10-17",
3
    "Statement": [
4
        {
5
            "Sid": "VisualEditor0",
6
            "Effect": "Allow",
7
            "Action": [
8
                "rds:StartDBCluster",
9
                "rds:StopDBCluster",
10
                "rds:ListTagsForResource",
11
                "rds:DescribeDBInstances",
12
                "rds:StopDBInstance",
13
                "rds:DescribeDBClusters",
14
                "rds:StartDBInstance"
15
            ],
16
            "Resource": "*"
17
        }
18
    ]
19
}

Input JSON Policy

Enter a name and click Next

Name the Policy

Create Role

Next, click Create Role

Create IAM Role

Choose Lambda and click Next

Choose Lambda service

Select the Policy you just created and click AWSLambdaBasicExecutionRole, then click Next

Select Policies

Enter the Role name

Name the Role

Confirm that the selected Policies appear, then click Create

Confirm and Create Role

Configure RDS

Set RDS Tags

Go to the RDS page and click on the instance you want to shut down long-term

Select RDS instance

Click on the Tags field, then click Manage tags to add two tags: autostart=yes and autostop=yes

Add RDS Tags

Set RDS Maintenance Time

RDS needs to be in a running state to configure settings. Click Modify to configure

Modify RDS

Set the maintenance time at the end of the settings. Please note that this is UTC time, not Taiwan time. The duration is 0.5 hours. Apply the same settings to other machines that need to be shut down long-term. After setting, click Continue and choose to apply immediately for the changes to take effect

Set Maintenance Window

Create Lambda Function

Create Start-up Lambda

Go to Lambda Function and click Create

Create Lambda Function

Enter a name, choose Python 3.11, and select x86_64 for the architecture

Configure Lambda

Select the Role you just created and click Create

Select Role for Lambda

Paste the following code and click Deploy

Deploy Lambda Code

1
import boto3
2
rds = boto3.client('rds')
3

4
def lambda_handler(event, context):
5

6
    #Start DB Instances
7
    dbs = rds.describe_db_instances()
8
    for db in dbs['DBInstances']:
9
        #Check if DB instance stopped. Start it if eligible.
10
        if (db['DBInstanceStatus'] == 'stopped'):
11
            try:
12
               GetTags=rds.list_tags_for_resource(ResourceName=db['DBInstanceArn'])['TagList']
13
                for tags in GetTags:
14
                #if tag "autostart=yes" is set for instance, start it
15
                    if(tags['Key'] == 'autostart' and tags['Value'] == 'yes'):
16
                        result = rds.start_db_instance(DBInstanceIdentifier=db['DBInstanceIdentifier'])
17
                        print ("Starting instance: {0}.".format(db['DBInstanceIdentifier']))
18
            except Exception as e:
19
                print ("Cannot start instance {0}.".format(db['DBInstanceIdentifier']))
20
                print(e)
21

22

23
if __name__ == "__main__":
24
    lambda_handler(None, None)

Set the timeout to 10 seconds. Click Configuration -> General configuration -> Edit

Edit Lambda Settings

Set to 10 seconds and save

Set Timeout

Set the periodic start-up trigger time. Click Add Trigger

Add Trigger

Choose EventBridge
Create a new rule
Enter a rule name
Choose to use a cron expression
AWS officially recommends entering 30 minutes before the RDS maintenance time, but this example uses the same time as the RDS maintenance time, entering cron(0 18 ? * SUN *). The time zone is still UTC. You can click here to refer to cron expressions
Click Add

Configure Trigger

Create Shutdown Lambda

All operations are the same as the start-up process above. The shutdown script is as follows:

1
import boto3
2
rds = boto3.client('rds')
3

4
def lambda_handler(event, context):
5

6
    #Stop DB instances
7
    dbs = rds.describe_db_instances()
8
    for db in dbs['DBInstances']:
9
        #Check if DB instance is not already stopped
10
        if (db['DBInstanceStatus'] == 'available'):
11
            try:
12
               GetTags=rds.list_tags_for_resource(ResourceName=db['DBInstanceArn'])['TagList']
13
                for tags in GetTags:
14
                #if tag "autostop=yes" is set for instance, stop it
15
                    if(tags['Key'] == 'autostop' and tags['Value'] == 'yes'):
16
                        result = rds.stop_db_instance(DBInstanceIdentifier=db['DBInstanceIdentifier'])
17
                        print ("Stopping instance: {0}.".format(db['DBInstanceIdentifier']))
18
            except Exception as e:
19
                print ("Cannot stop instance {0}.".format(db['DBInstanceIdentifier']))
20
                print(e)
21

22
if __name__ == "__main__":
23
    lambda_handler(None, None)

Testing

Click Test to perform start-up or shutdown tests

Test Lambda Function