Back up a MongoDB database to AWS S3 through a simple AWS Lambda function by using the mongodump binary. Result is a ZIP archive with .bson and .metadata.json files for each collection.

For a MongoDB Atlas cluster database backup, specify the URI command option like this:

--uri "mongodb+srv://[user]:[pass]@[host]/[name]"

Adapted from alonhar/lambda-mongodb-s3-backup.

mongodump binary is version 100.1.1 (mongodb-database-tools-amazon2-x86_64-100.1.1)

1. Clone repository and run npm install
2. ZIP contents of the folder (not the folder itself)
3. Create an AWS Lambda function
   • Select 'Author from scratch', enter your function name and select Node.js 12.x
   • Choose an existing role or create a new one and make sure it has a policy with s3:PutObject and s3:ListBucket permissions for the S3 bucket that you want to back up to, as well as the AWSLambdaBasicExecutionRole policy
   • Upload the ZIP file (no need to upload it to S3 first, as it should be just below 10MB so that it can be uploaded directly in Lambda)
   • Set environment variables (see table below)
   • Increase timeout from 3sec to 30sec
   • Configure a trigger. For instance, with CloudWatch Event rules (or Amazon EventBridge as it is now called), you can set up a cron schedule

v1.2.0

• Updated mongodump from 4.0.5 to 100.1.1
• The Mongo connection string is no longer fixed at a certain format. You now have the flexibility to specify the URI or host in the format you need, as well as other mongodump command options.
• There is a new env variable through which you can specify the ZIP filename (it is no longer the database name)
• The uploaded archive is now automatically AES256 encrypted on S3
• The S3 storage class is now defined on the file level instead of the bucket level
• Replaced 'zip-a-folder' by 'archive'
• Updated docs