0.4.0 • Published 7 years ago

lambda-env-vars v0.4.0

Weekly downloads
5
License
ISC
Repository
github
Last release
7 years ago

Lambda Environment Variables

A package for decrypting Lambda environment variables encrypted by AWS KMS

Coverage Status Build Status

The purpose of this package is the easily decrypt and fetch environment variables in Lambda functions, using KMS for decryption. The package supports getting environment variables that have been encrypted in Lambda using a default service key, however the main purpose is for decrypting variables that were encrypted using a custom KMS key. For more information on Lambda environment variables and encryption keys, see the AWS Documentation.

Before implementing it is recommended you read the FAQs section

Contents

Usage

AWS config

When using encrypted environment variables you will need to create a KMS key in IAM and give usage permission to the role that your Lambda function has been assigned. You then need to configure your Lambda function to use the new KMS key by default. This can be found in the Lambda function under Configuration -> Advanced settings -> KMS key.

Add lambda-env-vars to your project

$ npm install --save lambda-env-vars

If you're using yarn:

$ yarn add lambda-env-vars

Lambda Handler Example

It's important that the instance of the class is defined outside the handler, in order to utilize global variable caching, to cut down on KMS decryption charges.

ES6 Example:

import LambdaEnvVars from 'lambda-env-vars';
const lambdaEnvVars = new LambdaEnvVars();

function handler(event, context, callback) {
  // Get an environment variable encrypted using a custom KMS key.
  lambdaEnvVars.getCustomDecryptedValue('envVarKey')
    .then((decryptedValue) => {
      doSomethingWithDecryptedValue(decryptedValue);
    });

  // Get an environment variable that uses a default service key.
  const simpleKeyVariable = lambdaEnvVars.getDefaultDecryptedValue('simpleEnvVarKey');
  doSomethingWithSimpleVariable(simpleKeyVariable);
}

export { handler };

ES5 Example:

var LambdaEnvVars = require('lambda-env-vars');
var lambdaEnvVars = new LambdaEnvVars.default();

exports.handler = (event, context, callback) => {
  lambdaEnvVars.getCustomDecryptedValue('testVariable')
    .then(doSomethingWithDecryptedValue);

  const simpleKeyVariable = lambdaEnvVars.getDefaultDecryptedValue('simpleEnvVarKey');
  doSomethingWithSimpleVariable(simpleKeyVariable);
};

API Reference

Setting default config parameters

The methods getCustomDecryptedValue and getCustomDecryptedValueList both accept a second parameter object to allow users to specify the location ('s3' or 'lambdaConfig') of the environment variables. These parameters can be defined by default by specifying them when the class is initially instanced. For example:

import LambdaEnvVars from 'lambda-env-vars';
const lambdaEnvVars = new LambdaEnvVars({
  location: 's3',
  s3Config: {
    bucketName: 'my-env-var-bucket',
    fileName: 'my-env-var-filename.json',
  },
});

The default parameters can be overridden by simple specifying them on function call:

const params = { location: 'lambdaConfig' };
lambdaEnvVars.getCustomDecryptedValue('variableName', params)
  .then((variableName) => {
    console.log(variableName);
    // variableName will be fetched from the lambdaConfig
  });

Config parameters

The default config parameters object specifies that the source of environment variables should be the 'lambdaConfig', this is where users can define lambda env variables in the AWS console. Here's what that object looks like:

{
  location: 'lambdaConfig',
  s3Config: {},
}

Optionally the source of environment variables can be a JSON file within an S3 bucket. This is good for situations where your Lambda function config exceeds 4KB and you still need to store more environment variables. Encryption where the Lambda function environment variables are in S3 should be achieved with SSE-S3 or SSE-KMS. I'd personally recommend setting up a bucket policy to stop uploads that aren't encrypted at rest, just to be safe. One further thing to consider is your Lambda function's role will need a policy to allow reads to the bucket the environment variable file is stored in. The parameters required when using S3 as the source of the variables should look like the following:

{
  location: 's3',
  s3Config: {
    bucketName: 'name-of-bucket',
    fileName: 'filename.json', // Name of the JSON file containing the env vars
  },
}

The available attributes of the params object are as follows:

NameTypeDefaultInfo
locationstring'lambdaConfig'Can be 'lambdaConfig' or 's3'. The location the environment variables are stored.
s3Configobject{}Required when location is equal to 's3'. Fields that specify the location of the env var JSON file
s3Config.bucketNamestringThe name of the bucket containing the env var file.
s3Config.fileNamestringThe file name of the env var JSON file.

Decrypt an environment variable that uses a custom KMS key

Uses KMS to decrypt the cipher text stored under the environment variable of the specified key name. Caches the decrypted variable in the global scope so it is only decrypted once per container, cutting down on KMS decryption costs.

lambdaEnvVars.getCustomDecryptedValue('envVarKey');

Parameters:

NameTypeDefaultInfo
variableNamestring''The key in process.env to which the variable is stored under.
configParamsobjectconfigParamsOptional. Information that allows the package to know where to get the variable from. For more info see the config params section

Returns a promise that resolves the decrypted value, or rejects an error if there were issues connecting to KMS or issues with the encrypted payload.

Decrypt multiple environment variables that use a custom KMS key

Like the single variable, uses KMS to decrypt the cipher text stored under keys in the process.env object (Lambda environment variables). Again, the decrypted values are cached in the global scope, so the variables are only encrypted once per Lambda container. Multiple environment variable keys can be specified and they will be returned as keys to an object where the values are decrypted.

lambdaEnvVars.getCustomDecryptedValueList(['envVarKey1', 'envVarKey2']);
// returns { envVarKey1: 'Decrypted variable', envVarKey2: 'Decrypted variable' }

Parameters:

NameTypeDefaultInfo
variableNamesArray[]Keys in process.env to which encrypted environment variables are stored under.
configParamsobjectconfigParamsOptional. Information that allows the package to know where to get the variables from. For more info see the config params section

Returns an object containing the decrypted values where the keys are the items specified in the params variableNames.

Get an environment variable decrypted using a default service key

Returns the variable stored under process.env for the specified key. Default service key encrypted variables are decrypted before the Lambda invocation meaning the decrypted value is already available under process.env.

const value = lambdaEnvVars.getDefaultDecryptedValue('envVarKey');

Parameters:

NameTypeDefaultInfo
variableNamestring''The key in process.env to which the variable is stored under.

Returns the string value of the environment variable. No decryption takes place in code as this is done before Lambda is called.

FAQs

My Lambda config exceeds 4KB. What do I do?

Lambda imposes a 4KB limit on function config, this is inclusive of environment variables. and by using a few encrypted environment variables it easy to quickly reach this limit. The way this package recommends using environment variables when the Lambda config reaches 4KB is by storing a file in S3 that is encrypted at rest and pulling this down on the first invocation of the Lambda function and caching it for reduced calls to S3. For using environment variables stored in s3, see the config parameters section.

Why is the aws-sdk a dev dependency?

The package depends on the aws-sdk, however it is not listed as a dependency as it should be installed on your lambda environment by default.

Doesn't KMS decryption get quite expensive?

Yes, however as it is recommended in AWS's KMS helper code, the decrypted variables are stored in memory so only the first invocation of a Lambda function container incurs a KMS cost. All requests after this point will receive the var stored in memory.

Contributing

  • Start a feature branch from master
  • Tests should be written in the test directory for any new features.
  • Code should follow the installed style guide of airbnb.
  • Tests and linting can be run with npm test.
  • Once your feature is complete submit a PR to the master branch.