Backup MongoDB databases using AWS Lambda functions

The code in this repo allows running the backup of a MongoDB database to S3 on AWS Lambda by emulating a mongodump using pymongo and boto3. By default the function will create a JSON file for every collection in the database and save them in a /backup folder.

pip install mongo_lambda_backup

All configuration values are passed using environment variables.

The following values are required:

• MONGO_URI: The URI (including authentication credentials if needed) of the MongoDB host to be backed up. In case this contains credentials you can encrypt this value using AWS KMS and set MONGO_URI_IS_ENCRYPTED to tell the Lambda it needs to decrypt the given value.
• BUCKET_NAME: The name of the S3 bucket to store the backup files in.

The following values are optional:

• MONGO_URI_IS_ENCRYPTED: In case this environment variable is set, the handler assumes the URI (which potentially contains credentials) needs to be encrypted using the KMS key associated with the Lambda.
• MONGO_DATABASE: The name of the database to back up. In case this is not set, it will be read from the path of MONGO_URI.
• BUCKET_FOLDER: The folder in the bucket to store the JSON files in. Defaults to backups.
• COLLECTION_BLACKLIST: A comma-separated collection of collection names to skip when performing the back up (e.g. for skipping indices or vendor specific databases).
• IN_MEMORY: If set, the collection to be dumped to S3 will be buffered solely in the Lambda's memory before writing (instead of being temporarily saved to disk). This might affect both backup speed as well as Lambda Limits in certain scenarios.

MIT © Frederik Ring