@seagull/deploy-aws

Code deploying for the Seagull Framework

Usage no npm install needed!

<script type="module">
  import seagullDeployAws from 'https://cdn.skypack.dev/@seagull/deploy-aws';
</script>

README

@seagull/deploy

This package deploys a bundle that was handed over via cdk.

How to deploy something

For a generic solution you can just start the seagull-deploycommand. This will create a universal lambda, that boots an express server. You can handover the following parameters by env variables:

AWS_ACCOUNT_ID

  • default: undefined
  • the account id that will be used for this deployment
  • currently it is only used to determine the item-s3-bucket name, which must be uniq.
  • this parameter was added to test the app template without mocking AWS-STS in aws.sdk
  • best advice: just don't use it

BRANCH_NAME

  • default: master
  • the branch that will be deployed
  • it is basically used for the pipeline to create it's own project instead of overriding the prod deployment

DEPLOY_MODE

  • default: prod
  • differ whether the app should be deployed in test or prod mode
  • prod will use its own bucket, the test instances will use their own
  • non-prodapp will get a suffix BRANCH_NAME-DEPLOY_MODE added to the app name

AWS_PROFILE

  • default: default
  • the aws profile that is used for the deployment
  • should be configured on the ~/.aws/credentialsfile

AWS_REGION

  • default: eu-central-1
  • the region the app will be deployed
  • notice that there are some services in AWS that are not available everywhere

Advanced Deployment

  • you can use the basic seagull_app to have an app and use the convenience functions to add resources or use cdk itself to tweak the deployment. After that you can use the meth deployStack on the app to bring it to aws

How to see potential changes

To diff the current project in your repository with the one in aws just run seagull-diff. It uses the same env variables and will create an app like in seagull-deploy but does not put it into aws, instead it uses cloudformation-diff to show you what changed in your structure. Like the deployment you can also build an app yourself and that use the diffStack method to see changes.

How to create a Code Pipeline for something

To implement any kind of CI you can create a AWS Pipeline with seagull. Just use seagull-pipelineto create one. You can also create a custom one as its own seagull app. Just use the SeagullApp class for that. The default pipeline will use a github repository and creates a webhook to trigger itself. This means a new push to the watched branch will trigger a new build. therefore you will need to provide the data for the repositories like the owner, the repository name and a oAuthToken. The Token can also be stored in AWS-SSM, then you just have to forward the GITHUB_SSM_PARAMETER. Within the pipeline script a seagull-deploy will be executed. All parameters with the same name are given right trough to this deploy command,

seagull-pipeline will use the following env variables as options:

BRANCH_NAME

  • default: master
  • the name of the current branch, like in seagull-deploy
  • will be given to the deploy command within the pipeline

DEPLOY_MODE

  • default: prod
  • differ whether the app should be deployed in test or prod mode
  • prod will use its own bucket, the test instances will use their own

GITHUB_OWNER

  • default: undefined
  • the owner of the repository the pipeline should be triggered by

AWS_REGION

  • default: eu-central-1
  • the aws region the deployment should take place

GITHUB_REPO

  • default: undefined
  • the repository name that should be used as github repo for the pipeline

GITHUB_SSM_PARAMETER

  • default: undefined
  • the name of the OAuthToken in SSM

GITHUB_OAUTH

  • default: undefined
  • the oAuthToken to get access to the github repository

COMPUTE_TYPE_SIZE

  • default: SMALL
  • the ComputeType to use for the build environment (see CodeBuild docs)