Deploying Lambda functions manually involves quite a few steps.
Manually clicking buttons to upload zip files is fine the first few times
but gets old pretty quickly.
There is an easier way!
Simplify the process of deploying a AWS Lambda Function without having to adopt a build tool/system.
There are 5 Steps to setup deployment for your Lambda Function:
npm install dpl --save-dev
In order to use dpl
to deploy your Lambda function,
you need to have a few environment variables defined.
e.g: AWS_REGION
and AWS_IAM_ROLE
.
We have created a
.env_sample
file that shows you exactly which environment variables you need.
Simply copy it into your project. e.g:
cp node_modules/dpl/.env_sample .env && echo ".env\n" >> .gitignore
And then update the values with the real ones for your project.
Note: You need to have your AWS Credentials set to use the
aws-sdk
if you have not yet done this, see below for instructions.
In your package.json
file, add the list of files & directories
you want to be included in this zip that gets uploaded to AWS.
Example:
"files_to_deploy": [ "package.json", "index.js", "lib/" ]
Sample: package.json#L14-L17
Example:
"scripts": {
"deploy": "dpl"
}
npm run deploy
If you see an error message in your console, read the message and resolve it by correcting your setup. you have either not set your AWS Credentials or not defined the required environment variables. If you get stuck or have questions, ping us!
Deploying your Lambda function requires a few Environment Variables to be set.
As with all node.js code which uses the aws-sdk
,
it expects to have your AWS credentials stored locally.
Your credentials are expected to be at: ~/.aws/credentials
(e.g: if your username is alex, your AWS credentials will
be stored at /Users/alex/.aws/credentials
)
If you have not yet set your AWS credentials on your machine
do this now. Read about the AWS credential format.
The script needs to know which AWS_IAM_ROLE
you want to use to deploy/run the function.
Example:
export AWS_IAM_ROLE=arn:aws:iam::123456789:role/LambdaExecRole
Example:
export AWS_REGION=eu-west-1
To make the deployment script's run faster,
and avoid forcing people to add entries into their .gitignore
file,
we store the /dist
directory and resulting .zip
file
in your OS's Temporary storage.
e.g:
export TMPDIR=/path/to/where/you/want/dist/
For the deploy scrip to do its job,
we need to add two lines to our package.json
In your package.json
file, add the list of files & directories
you want to be included in your distribution.
Example:
"files_to_deploy": [ "package.json", "index.js", "lib/", ".env" ]
This tells dpl
to copy these files and directory (with all contents)
to the /dist
which will be zipped and sent to AWS.
Check our package.json#L14-L17 file if in doubt.
Instead of uploading all the files in a project to S3/Lambda we upload only
the required files. e.g: /src
or /lib
and ./index.js
.
While we are preparing this package, these required files are copied to
the (temporary) /dist
directory which will be zipped in Step 5.
Why a temporary directory? see: http://stackoverflow.com/questions/17946360/what-are-the-benefits-of-using-the-official-temp-directory-for-the-os
This typically includes the following:
lib/
- the directory containing custom code your lambda function uses.package.json
- the "manifest" file for your project, includes the Lambda function name, any configuration and dependencies.index.js
- the "main"handler
of your Lambda function.
We only need the "production" dependencies to be zipped and shipped.
so instead of copying all the devDependencies in node_modules
,
we simply install a fresh set using the --production
flag.
Once the /dist
directory has been created with the necessary files
and the dependencies have been installed in /dist/node_modules
we zip
the "distribution" ready for uploading to AWS.
This can take a few seconds depending on how many dependencies your Lambda function has.
Once the zip
has been packaged we upload it to AWS using the aws-sdk
.
Your Lambda function will be named according to the "name"
in
the package.json
file for your project.
Note: We are using the latest stable version of Node.js when deploying packages. see: https://aws.amazon.com/blogs/compute/node-js-4-3-2-runtime-now-available-on-lambda/
If for any reason you prefer/need to use the "Old" (legacy) version, please let us know: #33
-
Minial Dependencies - Our solution to the deployment task uses only one core dependency: the
aws-sdk
. -
Small Code - The entire
dpl
("Deploy Lambda") module is fewer lines than our original
gulpfile.js
and uses only node.js core modules (you know and love) and your OS-nativezip
command. -
A beginner can read and understand all the code used in
dpl
in a few minutes; our code has both JavaDoc and in-line comments and we are here to help if you have any questions! -
No assumptions about your code style. e.g if you need any custom processing simply add a task in your
scripts
section of yourpackage.json
and run that task before deploying. -
No Global packages required or implied, just one
dev
Dependency.
###Β Optional configuration
Want to specify the MemorySize
or Timeout
settings for your Lambda function?
"lambda_memory"
- maximum memory allocation for your lambda function."lambda_timeout"
- maximum execution time.
In your package.json
add:
"lambda_memory":"512",
"lambda_timeout": "30",
Unlike other AWS Lambda deployment methods,
dpl
lets you use environment
variables in your Lambda Functions!
Simply add .env
to your list of "files_to_deploy"
in your package.json
Example:
"files_to_deploy": [ "package.json", "index.js", "lib/", ".env" ]
And an .env
file will included in your .zip
file that gets uploaded to AWS.
This means you can use an Environment Variable loader
e.g env2
in your Lambda function:
require('env2')('.env');
exports.handler = (event, context) => {
if (event.hello === 'world') {
return context.fail(JSON.stringify({
statusCode: 500,
message: 'sorry'
}));
} else {
return context.succeed({
statusCode: 200,
message: process.env.HELLO // or how ever you use environment variables!
});
}
};
- Using NPM as a build tool: http://blog.keithcirkel.co.uk/how-to-use-npm-as-a-build-tool/ (you don't need gulp/grunt/etc...)