cdk-gradle-uploader 0.0.39

Uploads new Gradle versions to an S3 bucket

Gradle Uploader

This CDK construct checks for new releases of the Gradle build software.

The new release will be made available as copy in an S3 bucket. An information about the new release can be sent via e-mail or via Slack.

Internally the construct uses

• an S3 bucket for storing the Gradle software

• a Lambda function and one Lambda layer to

  • check for the latest Gradle release
  • upload if required and notify users via SNS and e-Mail or alternatively Slack

• a Cloudwatch event rule to trigger the Lambda function

Setup of the components

The S3 Bucket

By default, public access to the S3 bucket is disabled. Only the access from a specific IP address (the one I got from my ISP) is allowed and ensured via bucket policies.

  const bucket = new Bucket(this, "bucket", {
      blockPublicAccess: BlockPublicAccess.BLOCK_ALL,
      encryption: BucketEncryption.S3_MANAGED,
      publicReadAccess: false,
      versioned: false,
      removalPolicy: RemovalPolicy.DESTROY,
    });

    const bucketContentStatement = new PolicyStatement({
      effect: Effect.ALLOW,
      actions: ["s3:GetObject"],
      resources: [bucket.bucketArn + "/*"],
      principals: [new AnyPrincipal()],
    });
    bucketContentStatement.addCondition("IpAddress", {
      "aws:SourceIp": "87.122.220.125/32",
    });

    const bucketStatement: PolicyStatement = new PolicyStatement({
      effect: Effect.ALLOW,
      actions: ["s3:ListBucket", "s3:GetBucketLocation"],
      resources: [bucket.bucketArn],
      principals: [new AnyPrincipal()],
    });
    bucketStatement.addCondition("IpAddress", {
      "aws:SourceIp": "87.122.220.125/32",
    });

    const bucketPolicy = new BucketPolicy(this, "bucketPolicy", {
      bucket: bucket,
    });

The Lambda function

The Lambda function is written in Python (version 3.8). The execution time is limited to five minutes and the memory consumption to 512 MByte. Additionally the function gets read/ write access to the S3 bucket and has a log retention period is set to one week.

const fn = new Function(this, "fnUpload", {
  runtime: Runtime.PYTHON_3_8,
  description: "Download Gradle distribution to S3 bucket",
  handler: "gradleUploader.main",
  code: Code.fromAsset("./lambda/"),
  timeout: Duration.minutes(5),
  memorySize: 512,
  logRetention: RetentionDays.ONE_WEEK,
  layers: [layer],
  environment: {
    BUCKET_NAME: bucket.bucketName,
    TOPIC_ARN: topic.topicArn,
  },
});

bucket.grantReadWrite(fn);

If Slack is selected as notification channel, then also the WEBHOOK_URL is part of the environment.

In the additional layer modules like boto3 are included.

const layer = new LayerVersion(this, "GradleUploaderLayer", {
  code: Code.fromAsset(path.join(__dirname, "../layer-code")),
  compatibleRuntimes: [Runtime.PYTHON_3_8],
  license: "Apache-2.0",
  description: "A layer containing dependencies for thr Gradle Uploader",
});

The Cloudwatch event rule

Every first of a month the Lambda function fn will be triggered automatically. That seems to be a reasonable period for the update check.

const target = new LambdaFunction(fn);
new Rule(this, "ScheduleRule", {
  schedule: Schedule.cron({ minute: "0", hour: "0", day: "1", month: "*" }),
   targets: [target],
});

Notifying about new releases

Whenever the release of a new Gradle version is detected, the stack will sent an e-mail to the list of subscriber using SNS.

private addSubscribers(topic: Topic, subscribers:Array<string>) {
    for (var subscriber of subscribers) {
      topic.addSubscription(new EmailSubscription(subscriber));
    }
  }

The forwarding of information to a Slack channel is done from within the Lambda function.

Testing the Python code

docker run --rm -v "$PWD":/var/task:ro,delegated   -v /home/stefan/Private/programmieren/aws/cdk/gradle_uploader/layer-code:/opt:ro,delegated  -e AWS_ACCESS_KEY_ID=XXXXXXXXXX -e AWS_SECRET_ACCESS_KEY=XXXXXXXXXX lambci/lambda:python3.8 gradleUploader.main

How to use the construct in a stack

Here is an example how to use the construct:

export class GradleUploaderStack extends Stack {
  constructor(scope: Construct, id: string) {
    super(scope, id);
    new GradleUploader(this, 'TestStack', {
      mailProperties: { subscribers: ['<e-mail address>'] },
      slackProperties: {
        webhook:
          'https://hooks.slack.com/services/T00000000/B00000000/XXXXXXXXXXXXXXXXXXXXXXXX',
      },
      whitelist: ['CIDR_1', 'CIDR_2'],
    });
  }
}

const app = new App();
new GradleUploaderStack(app, 'TestApp');
app.synth();

Links

• AWS Cloud Development Kit
• Gradle Homepage
• boto3
• Slack