RDS Snapshot All

A simple Python script to create RDS snapshots across all AWS regions in your account.
Runs directly in AWS CloudShell with no external dependencies beyond boto3.

Features

• Scans all opted-in AWS regions for RDS instances

• Creates DB snapshots for standalone RDS instances

• Creates cluster snapshots once per Aurora cluster

• Skips read replicas by default (optional flag to include them)

• Auto-tags snapshots with CreatedBy=CloudShellScript and account ID

• Snapshot names are timestamped to avoid collisions

• Prints a summary of created, skipped, and errored snapshots

Usage

1. Open AWS CloudShell in your AWS account.

2. Clone this repository or copy the script:

   git clone https://github.com/veup-engineering/rds-snapshotter.git

3. Change into the RDS Snapshot directory:

cd rds-snapshotter

4. Create a virtual environment:

   python3 -m venv rds-env

5. Activate the virtual environment:

   source rds-env/bin/activate

6. Install boto3:

   pip install boto3

7. Run the script:

   python3 rds_snap_all.py

Examples

Snapshot all RDS instances and Aurora clusters across all regions:

python3 rds_snap_all.py

Snapshot only in us-east-1 and ca-central-1:

python3 rds_snap_all.py --regions us-east-1,ca-central-1

Include read replicas:

python3 rds_snap_all.py --include-replicas

Custom snapshot prefix and tags:

python3 rds_snap_all.py --prefix nightly --tag Env=Prod --tag Owner=DevOps

IAM Permissions

Your CloudShell user/role must allow these actions:

• rds:DescribeDBInstances

• rds:CreateDBSnapshot

• rds:CreateDBClusterSnapshot

• rds:AddTagsToResource

• ec2:DescribeRegions

• sts:GetCallerIdentity

Example inline policy:

{
  "Version": "2012-10-17",
  "Statement": [{
    "Effect": "Allow",
    "Action": [
      "rds:DescribeDBInstances",
      "rds:CreateDBSnapshot",
      "rds:CreateDBClusterSnapshot",
      "rds:AddTagsToResource",
      "ec2:DescribeRegions",
      "sts:GetCallerIdentity"
    ],
    "Resource": "*"
  }]
}

Notes

• The script does not wait for snapshots to finish, it only initiates them.

• Aurora clusters: only one cluster snapshot per DBClusterIdentifier is created.

• Read replicas are skipped unless you pass --include-replicas.

• Snapshot IDs are timestamped in UTC, e.g. autosnap-db-mydb-20250909-053000.

• Tags are attached both at snapshot creation and via AddTagsToResource for consistency.