Overseer

A URL monitoring service running 100% serverless in AWS.

Requirements

• AWS Account
• AWS CLI installed and configured
  • Default profile will be used when deploying
• Docker
• Node 14.x

Configuration

Add a .env file to the root of the project with the following variables declared:

// For alerting
DISCORD_WEBHOOK_URL= // the discord webhook url that you would like to use for notifications. (temporary)

// For website hosting and routing
HOSTED_ZONE_ID=yourroute53hostedzoneid
DOMAIN_NAME=yourwebsite.com
DASHBOARD_DOMAIN=dashboard.yourwebsite.com // MUST be a subdomain of DOMAIN_NAME
ACM_CERT_ARN=some:aws:arn // this cert should be issued for your domain name

// for authentication via AWS cognito
AMAZON_CLIENT_ID=youramazonoauthclientid
AMAZON_CLIENT_SECRET=youramazonoauthsecret
GOOGLE_CLIENT_ID=yourgoogleoauthclientid
GOOGLE_CLIENT_SECRET=yourgoogleoauthsecret

Doing Stuff

yarn start # starts proxy and local development server

yarn deploy # builds and deploys stack to AWS

Notes

• AFTER first deployment, you will need to create a config.json file in the root of the /dashboard directory.
• The config.json file should contain the following:

{
  "userPoolId": "the-user-pool-id",
  "userPoolClientId": "the-user-pool-client-id",
  "authDomain": "auth.yourwebsite.com"
}

• You will also have to add the auth.yourwebsite.com domain to the cognito custom domain configuration, and add the cloudfront distribution it gives as a result to the route53 hosted zone as a CNAME.

Stack deployment issues:

• UserPoolDomain seems to get upset about root domain not resolving. Currently have to deploy with random cognito domain first, then manually add custom domain in console and update Route53 record.
• This is obviously not ideal. I think it can be resolved by splitting up the app into multiple stacks, or by taking another look at inter stack dependencies.