Flyway AWS Lambda Function

Run Flyway database migrations as a Lambda function.

ℹ️ This fork of the (seemingly abandoned) flyway-awslambda project introduces
the following improvements:

• Flyway can be configured via the officially supported environment variables (only config file was supported before)
• All Flyway configuration options are supported (only a small subset was supported before)
• Upgrade Flyway to version 6 (was 4)
• Improved documentation and examples

Other than possible breaking changes between Flyway 4 vs. 6 (I couldn’t identify any) the Lambda function itself is
100% backwards compatible with the original project, i.e. the JAR can be put in place of the old function without breaking changes.

How it works

1. The Lambda function is triggered either
   • by invoking it manually
   • or automatically when a file is uploaded into an S3 bucket
2. The migration scripts are loaded from the bucket
3. The migrations are performed on the database
4. A file containing the migration results is pushed to the bucket

Setup

ℹ️ CloudFormation samples can be found under ./src/main/aws/. Read on for the manual setup.

First, create a new Lambda function and upload the flyway-awslambda-x.y.z.jar file as code. You can either download a pre-built
JAR file from the releases or build it yourself (Scala is required):

sbt assembly

Settings

• Runtime: Java 8
• Handler:
  • “crossroad0201.aws.flywaylambda.S3EventMigrationHandler” if the Lambda should be triggered when a file is uploaded to the bucket
  • “crossroad0201.aws.flywaylambda.InvokeMigrationHandler” if you wish to invoke the Lambda manually
• Role:
  • AmazonRDSFullAccess
  • AmazonS3FullAccess
  • AWSLambdaVPCAccessExecutionRole
• VPC: Same VPC as the target RDS

Triggers (in case of S3EventMigrationHandler)

Bucket Structure

The Lambda function expects the S3 bucket to be structured like this:

s3://my-flyway             <- Flyway migration bucket.
  - /my-application        <- Flyway resource folder(prefix).
    - flyway.conf          <- Flyway configuration file (optional)
    - V1__create_foo.sql   <- SQL file(s)
    - V2__create_bar.sql

Running the Lambda

Using the S3EventMigrationHandler

• Put the Flyway SQL files into the migration folder (one by one!).
• The Lambda function is invoked automatically by an S3 event.
• Check migration-result.json in the S3 bucket for the results or CloudWatch logs for more detail.

Using the InvokeMigrationHandler

• Put the Flyway SQL files into the migration folder
• Invoke the Lambda function manually with the following JSON payload.
  (invoke by AWS console, CLI, any application…etc. see CLI example)

{
  "bucket_name": "my-flyway",
  "prefix": "my-application",
  "flyway_conf": "flyway.conf"  // Optional. Default 'flyway.conf'
}

• Check migration-result.json in the S3 bucket for the results or CloudWatch logs for more detail.

Configuration

The Lambda function can be configured either by way of a Flyway configuration file in the S3 bucket (1) or with
environment variables (2). In case no configuration file is found it is assumed that the function should be configured
via environment variables. Please note that the configuration method is a binary choice — you can only pick one.

1. Configuration file: For this configuration method place a Flyway compliant config file
   in the application folder along with the migration scrips.

2. Environment variables: To configure the Lambda function via env vars it is crucial that no config
   file is present in the bucket. The Lambda function passes the FLYWAY_* env vars directly through to the library. See
   the official documentation for a list of supported env vars.

Supported Databases

Only MySQL is supported out of the box. If another database is used it is recommended to make use of Lambda Layers.

Here’s how to:

• Download the desired database driver’s JAR file
• Zip the JAR such that the jar is located in the java/lib folder, e.g. java/lib/postgresql-42.2.6.jre7
• Create the Lambda layer (select the “Java 8” runtime) and assign it to the Lambda

An example CloudFormation template demonstrating a PostgreSQL setup via Lambda layers can be found in /src/main/aws/3-postgres.yaml.