yarn add @aws-abaschen/cdk-typescript
mkdir -p src/{functions,layers}
Refer to this example Stack and the related CDK App
This construct will allow your lambdas to output .mjs
files with their source maps. It relies on standardized folder structure in src/
Each Lambda function will have to be written in a folder which will become its functionName
in src/functions/{functionName}/index.ts
.
All the default values are for example purpose and doesn't reflect any recommendation nor will be maintained.
Instead of letting CDK generating a role, the construct will create one named fnRole${id}
and will append the policies below with a Stack id fn-role-${id}
. The property role
is also available for further manipulation.
By default, the PolicyStatement to write in the log group will be added:
new PolicyStatement(
{
effect: Effect.ALLOW,
actions: [
"logs:CreateLogGroup",
"logs:CreateLogStream",
"logs:PutLogEvents",
],
resources: [`arn:aws:logs:${Stack.of(this).region}:${scope.account}:log-group:/aws/lambda/${id}:*`]
}
);
Adding a VPC to your lambda function will automatically add the policy to connect:
new PolicyStatement({
effect: Effect.ALLOW,
actions: [
"ec2:CreateNetworkInterface",
"ec2:DescribeNetworkInterfaces",
"ec2:DeleteNetworkInterface",
"ec2:AssignPrivateIpAddresses",
"ec2:UnassignPrivateIpAddresses"
],
resources: ["*"]
});
Tracing is enabled by default and the policy statement will be added unless disabled:
new PolicyStatement({
effect: Effect.ALLOW,
actions: [
"logs:CreateLogDelivery",
"logs:DeleteLogDelivery",
"logs:DescribeLogGroups",
"logs:DescribeResourcePolicies",
"logs:GetLogDelivery",
"logs:ListLogDeliveries",
"logs:PutResourcePolicy",
"logs:UpdateLogDelivery",
"xray:GetSamplingRules",
"xray:GetSamplingTargets",
"xray:PutTelemetryRecords",
"xray:PutTraceSegments"
],
resources: [`arn:aws:logs:${Stack.of(this).region}:${scope.account}:log-group:/aws/lambda/${id}:*`],
})
String parameters can be automatically referenced and read policy applied if they follow a specific prefix (default here is fn-
). They will then get injected as an environment variable with PARAM_
prefix and -
(dash) replaced with _
(underscore) in the name:
new NodetsFunction(this, name, {
parameters: ['DB-USER'],
environment: {
TEST_1: 'test Value 1'
}
});
Will output:
[...]
Environment:
Variables:
NODE_OPTIONS: --enable-source-maps
PARAM_DB_USER:
Ref: discordpongfnparamDBUSERParameter70CB30BD
TEST_1: test Value 1
AWS_NODEJS_CONNECTION_REUSE_ENABLED: "1"
[...]
Parameters:
discordpongfnparamDBUSERParameter70CB30BD:
Type: AWS::SSM::Parameter::Value<String>
Default: fn-DB-USER
Similarly, a layer will be named and the source has to be in a folder layerName
in src/layers/{layerName}/index.ts
.
Then a layer has to be referenced with import { someExport } from '@layer/layerName'
. The aliasing for typescript is in tsconfig.json $.aliases
and $.compilerOptions.paths
.
Example in Discord authorizer layer:
export const verify = async (event: APIGatewayEvent): Promise<Boolean> => {
...
}
Can be referenced in the lambda function discrod-pong:
import { verify } from '@layer/discord-authorizer';
Explanations:
During layer build, the layer is compiled into a module with a package.json
file. At runtime this module will end up in the lambda filesystem inside the /nodejs/node_modules/@layer\/discord-authorizer
. Thus when referencing '@layer/discord-authorizer` it will automatically find it.
The ResourceAspect
will tag every Cloud Formation resource with whatever static tag given in the props:
const fn = new NodetsFunction(this, 'fn', {...});
Aspects.of(fn).add(new ResourceAspect({
app: 'SampleApp'
}));
Will result in the template:
fnreturn2005A3631C2:
Type: AWS::Lambda::Function
Properties:
# ...
Tags:
- Key: resource:type
Value: AWS::Lambda::Function
- Key: x-app
Value: SampleApp
Any static tag is prefixed with x-
. The aspect can also be applied to the App
or Stack
directly for recursive tagging. Some resource may be missed like the lambda for log retention in a lambda function.