Hands down the best way to do this in codepipeline is by using https://serverless.com/ framework. This replaces every super complicated hack-job and workaround I've previously used. Way less complicated IMO.
Create a codepipeline, link it to src & a codebuild project, set a few permissions, done.
//serverless.yml
service: my-api
provider:
name: aws
runtime: python2.7
functions:
hello:
handler: handler.hello
events:
- http:
path: api/v1/message
method: post
//buildspec.yml
version: 0.2
phases:
install:
commands:
#BUILD
- sudo apt-get update -y
build:
commands:
- echo $environment
- serverless package --stage $environment --region us-east-1
- serverless deploy --stage $environment --region us-east-1
Or torture yourself by doing one of the options below...
You can do this in cloudformation from within code pipeline. Export the swagger spec from within the gatewayapi console and place in the cloudformation template.
AWSTemplateFormatVersion: '2010-09-09'
Resources:
PlayersAPI:
Type: AWS::ApiGateway::RestApi
Properties:
Name: MyApi
Description: API Description
Body:
"SWAGGER HERE"
Hooking this up to lambda is a little bit cumbersome but I can describe the steps. First create a codepipeline project with source, build, and deploy steps.
- src should be standard from github or codecommit
- build should output a zip file and use buildspec.yml Something like this...
//buildspec.yml
version: 0.1
phases:
install:
commands:
#BUILD
- zip -r lambda.zip . -x *.git*
artifacts:
files:
- '**/*.zip'
- '**/*.yml'
discard-paths: no
Have the build step export an artifact MyAppBuild (or whatever you want to call it)
The final pipeline step is creating the lambda function in this repo as a standalone function through the console(its reusable):
https://github.com/tkntobfrk/codepipeline-lambda-s3
This lambda function downloads the pipeline artifact/zipped lambda function and updates it using boto.
After these steps you could add another step as a cloudformation deploy step. Connect it to the lambda function that you've just deployed.
If you are dealing with multiple environments you could create lambda functions and gatewayapi cloudformation template for each environment then run them in sequence.
- stage 1: src
- stage 2: build
- stage 3: deploy lambda test, deploy gateway api cloudformation test
- stage 4: validate test
- stage 5: deploy lambda prod, deploy gateway api cloudformation prod
Using straight AWS serverless like this works too. However, you need to use a standard artifact location for the uri's. The DefinitionUri: for the API can be the exported swagger from the gatewayapi console.
//cloudformation.yml
AWSTemplateFormatVersion: '2010-09-09'
Transform: AWS::Serverless-2016-10-31
Resources:
MySimpleFunction:
Type: AWS::Serverless::Function
Properties:
Handler: app.lambda_handler
Runtime: python2.7
CodeUri: s3://somebucket/somezip.zip
MyAPI:
Type: AWS::Serverless::Api
Properties:
StageName: prod
DefinitionUri: s3://somebucket/somezip.zip
