Serverless Framework & Lambda: A Match Made in Heaven

One thing that I have always been keen to get from AWS Lambda is configurable environment variables that can be set for each individual Lambda function. One thing I loved about the Serverless framework is that it supports a notion of environment variables, stored in a hidden ‘_meta’ folder and substituted in when the functions are deployed to Lambda.

On our serverless-hipchat-connect project, we added persistence using DynamoDB, a serverless NoSQL database. To create this database, we added it to the `s-resources-cf.json`, such that the required tables would be created as part of a stage’s initialization.

The cool feature in this is that the Serverless Framework can capture the outputs of a CloudFormation template as environment variables for Lambda functions!

"DynamoDBAccessTokenTable": {
      "Type": "AWS::DynamoDB::Table",
      "DependsOn": "DynamoDBInstallationTable",
      "Properties": {
      "AttributeDefinitions": [
        {
            "AttributeName": "oauthId",
            "AttributeType": "S"
        }
      ],
        "KeySchema": [
        {
            "AttributeName": "oauthId",
            "KeyType": "HASH"
        }
      ],
        "ProvisionedThroughput": {
          "ReadCapacityUnits": 3,
          "WriteCapacityUnits": 1
      }
    }
  }
  },
  "Outputs": {
  "AccessTokenTableName": {
      "Description": "Name of the AccessTokenTable created by CloudFormation",
      "Value": {
        "Ref": "DynamoDBAccessTokenTable"
    }
  }
  }

This can then be referenced in a CloudFormation template’s configuration file:

  "environment": {
    "SERVERLESS_PROJECT": "${project}",
    "SERVERLESS_STAGE": "${stage}",
    "SERVERLESS_REGION": "${region}",
    "ACCESS_TOKEN_TABLE": "${accessTokenTableName}"
  },

At GorillaStack, we like to give a bit back to the community. If you like what you see here you should check out the rest of our Open Source DevOps contributions.