insta-integration - Integration Testing

Automated integration tests for any application/job.

• Spin up any external services
• Generate production-like data
• Run data validations to ensure application/job works as expected

Problems it can help with:

• Unreliable test environments
• Dependencies on other teams
• Simulate complex data flows

Usage

CLI

1. Install via npm install -g insta-integration

2. Create YAML file insta-integration.yaml to define your integration tests

   1. Examples can be found here.
   2. Use JSON schema to help guide you on available options

3. Run insta-integration

GitHub Action

1. Create YAML file .github/workflows/integration-test.yaml

   name: Integration Test
   on:
     push:
       branches:
         - *
   jobs:
     integration-test:
       name: Integration Test
       runs-on: ubuntu-latest
       steps:
         - name: Run integration tests
           uses: data-catering/insta-integration@v1

2. Create YAML file insta-integration.yaml to define your integration tests

   1. Examples can be found here.
   2. Use JSON schema to help guide you on available options

3. Push your code and the GitHub Action will run

Generation and Validation

Since it uses data-caterer behind the scenes to help with data generation and validation, check the following pages for discovering what options are available.

• Data Generation
• Data Validation

Data Sources

The following data sources are available to generate/validate data.

Full Example

services:
  - name: postgres #define external services
    data: my-data/sql #initial service setup (i.e. schema/tables, topics, queues)
run:
  - command: ./my-app/run-postgres-extract-app.sh #how to run your application/job
    env: #environment variables for your application/job
      POSTGRES_URL: jdbc:postgresql://postgres:5432/docker
    test:
      env: #environment variables for data generation/validation
        POSTGRES_URL: jdbc:postgresql://postgres:5432/docker
      mount: #volume mount for data validation
        - ${PWD}/example/my-app/shared/generated:/opt/app/shared/generated
      relationship: #generate data with same values used across different data sources
        postgres_balance.account_number: #ensure account_number in balance table exists when transaction created
          - postgres_transaction.account_number
      generation: #define data sources for data generation
        postgres:
          - name: postgres_transaction #give it a name to use in relationship definition
            options: #configuration on specific data source
              dbtable: account.transactions
            count: #how many records to generate (1,000 by default)
              perColumn: #generate 5 records per account_number
                columnNames: [account_number]
                count: 5
            schema: #schema of the data source
              fields:
                - name: account_number
                  type: string
                - name: create_time
                  type: timestamp
                - name: transaction_id
                  type: string
                - name: amount
                  type: double
          - name: postgres_balance
            options:
              dbtable: account.balances
            schema:
              fields:
                - name: account_number
                  options: #additional metadata for data generation
                    isUnique: true
                    regex: ACC[0-9]{10}
                - name: create_time
                  type: timestamp
                - name: account_status
                  type: string
                  options:
                    oneOf: [open, closed]
                - name: balance
                  type: double
      validation:
        csv: #define data source for data validations
          - options:
              path: /opt/app/shared/generated/balances.csv
              header: true
            validations: #list of validation to run, can be basic SQL, aggregations, upstream data source or column name validations
              - expr: ISNOTNULL(account_number)
              - aggType: count
                aggExpr: count == 1000
          - options:
              path: /opt/app/shared/generated/transactions.csv
              header: true
            validations:
              - expr: ISNOTNULL(account_number)
              - aggType: count
                aggExpr: count == 5000
              - groupByCols: [account_number]
                aggType: count
                aggExpr: count == 5

GitHub Action Options

Input

Optional configurations to alter the files and folders used by the GitHub Action can be found below.

To use these configurations, alter your .github/workflows/integration-test.yaml.

name: Integration Test
on:
  push:
    branches:
      - *
jobs:
  integration-test:
    name: Integration Test
    runs-on: ubuntu-latest
    steps:
      - name: Run integration tests
        uses: data-catering/insta-integration@v1
        with:
          configuration_file: my/custom/folder/insta-integration.yaml

Output

If you want to use the output of the GitHub Action, the following attributes are available:

For example, you can print out the results like below:

- name: Run integration tests
  id: test-action
  uses: data-catering/insta-integration@v1
- name: Print Output
  id: output
  run: |
    echo "Records generated:         ${{ steps.test-action.outputs.num_records_generated }}"
    echo "Successful validations:    ${{ steps.test-action.outputs.num_success_validations }}"
    echo "Failed validations:        ${{ steps.test-action.outputs.num_failed_validations }}"
    echo "Number of validations:     ${{ steps.test-action.outputs.num_validations }}"
    echo "Validation success rate:   ${{ steps.test-action.outputs.validation_success_rate }}"

JSON Schema for insta-integration.yaml

A JSON Schema has been created to help guide users on what is possible in the insta-integration.yaml. The links below show how you can import the schema in your favourite IDE:

• IntelliJ
• Visual Studio Code

Example Flows

Examples can be found here.

Why Integration Test?

It is the closest you get to simulating production. This involves:

• Test your application/job end-to-end
• Connect to data sources
• Production-like data processed