Setup

This setup explains how to start receiving budget notifications on slack. The Services used in this setup include Pub/Sub, Budgets, BigQuery, and Cloud Functions. A Budget message will be sent to Pub/Sub which will trigger a Cloud Function where the function uses the submitted Pub/Sub data to create an entry in a table within BigQuery. After data has been sent to BigQuery, a query and a condition take place to ensure that duplicate messages are not sent to Slack. Using a BigQuery table to place the data first allows for the Cloud Function to analyze the data before sending it to Slack.

1. Go to https://api.slack.com/apps and create an applicarion with the following permissions under OAuth and Permissions > Scopes > Bot Token Scopes.

   • chat:write.public
   • channels:join
   • chat:write
   • chat:write.customize

2. Save the Bot User OAuth Access Token somewhere as it is needed for the Cloud Function.

3. In BigQuery, add a new dataset called billing and create a new table called budget. The schema for the table is shown below:

[
 {
 "name": "createdAt",
 "type": "TIMESTAMP",
 "mode": "NULLABLE"
 },
 {
 "name": "costAmount",
 "type": "NUMERIC",
 "mode": "NULLABLE"
 },
 {
 "name": "budgetAmount",
 "type": "NUMERIC",
 "mode": "NULLABLE"
 },
 {
 "name": "budgetName",
 "type": "STRING",
 "mode": "NULLABLE"
 },
 {
 "name": "budgetId",
 "type": "STRING",
 "mode": "NULLABLE"
 },
 {
 "name": "threshold",
 "type": "NUMERIC",
 "mode": "NULLABLE"
 }
 ]

4. Create a new Cloud Pub/Sub topic called billing-alerts. This topic will be used for the billing alert publisher.

5. Navigate to Billing in the Google Cloud Console and create a budget. When creating your budget, modify "Manage notifications" to configure the Cloud Pub/Sub topic billing-alerts that was created in the previous step¹.

6. Navigate to Cloud Functions in the Google Cloud Console and create a new cloud function with the following settings:

   • Name: slack-billing-notification
   • Trigger: Cloud Pub/Sub
   • Topic: billing-alerts
   • Source code: inline editor
   • Runtime: Node.js 10
   • Function to execute: notifySlack

7. Configure two Runtime variables:

   • GCP_PROJECT: The project to be used for the BigQuery Dataset.
   • BOT_ACCESS_TOKEN: The Bot Token that was created in the first step.

8. Add the index.js and package.json files to the inline editor.

9. Click deploy.

Testing

It is requried to encode the Pub/Sub test data as Base64. The snippet below is some plain JSON data that would be similar to the actual data.

{
 "costAmount": 999,
 "budgetAmount": 1000,
 "budgetDisplayName": "test",
 "alertThresholdExceeded": 0.99
}

We need to take that plain JSON and convert it to Base64 format. You can do this by going to https://www.browserling.com/tools/json-to-base64. The result is this.

ewogImNvc3RBbW91bnQiOiA5OTksCiAiYnVkZ2V0QW1vdW50IjogMTAwMCwKICJidWRnZXREaXNwbGF5TmFtZSI6ICJ0ZXN0IiwKICJhbGVydFRocmVzaG9sZEV4Y2VlZGVkIjogMC45OQp9

Finally we put that output into the following JSON. Use this message as the test input for the Cloud Function.

{
 "data": "ewogImNvc3RBbW91bnQiOiA5OTksCiAiYnVkZ2V0QW1vdW50IjogMTAwMCwKICJidWRnZXREaXNwbGF5TmFtZSI6ICJ0ZXN0IiwKICJhbGVydFRocmVzaG9sZEV4Y2VlZGVkIjogMC45OQp9",
 "attributes": {
       "budgetId": "test"
 }
}

And that is all, you should see a notification when you run the test. The result should be...

I recommend adding an icon to the app to make it pretty. You can do that here: https://api.slack.com/apps.