monthly-dr-test.yml

name: Monthly DR Test

on:
  schedule:
    # Run at 06:00 UTC on the 1st of every month
    - cron: '0 6 1 * *'
  workflow_dispatch:
    inputs:
      dr_region:
        description: 'DR region to test'
        required: true
        default: 'us-west-2'
        type: string
      primary_region:
        description: 'Primary region'
        required: true
        default: 'us-east-1'
        type: string

env:
  PRIMARY_REGION: ${{ inputs.primary_region || 'us-east-1' }}
  DR_REGION: ${{ inputs.dr_region || 'us-west-2' }}
  NAMESPACE: gistpin

jobs:
  dr-test:
    name: Run DR Test
    runs-on: ubuntu-latest
    permissions:
      contents: read
      id-token: write

    steps:
      - name: Checkout code
        uses: actions/checkout@v4

      - name: Configure AWS credentials
        uses: aws-actions/configure-aws-credentials@v4
        with:
          role-to-assume: ${{ secrets.AWS_DR_TEST_ROLE_ARN }}
          aws-region: ${{ env.DR_REGION }}
          role-duration-seconds: 3600

      - name: Set up kubectl
        uses: azure/setup-kubectl@v4
        with:
          version: 'latest'

      - name: Run DR test script
        id: dr-test
        run: |
          bash infrastructure/scripts/dr-test-runner.sh \
            --primary-region "${PRIMARY_REGION}" \
            --dr-region "${DR_REGION}" \
            --namespace "${NAMESPACE}" \
            --rto 900 \
            --rpo 300
        continue-on-error: true

      - name: Archive DR test results
        if: always()
        uses: actions/upload-artifact@v4
        with:
          name: dr-test-results-${{ github.run_id }}
          path: /tmp/dr-test-runner-*.txt
          retention-days: 365

      - name: Upload DR test report
        if: always()
        uses: actions/upload-artifact@v4
        with:
          name: dr-test-report-${{ github.run_id }}
          path: /tmp/dr-test-runner-*.txt
          retention-days: 365

      - name: Notify on failure
        if: steps.dr-test.outcome == 'failure'
        uses: slackapi/slack-github-action@v1
        with:
          slack-message: "❌ *DR Test Failed* — ${{ github.repository }}/${{ github.run_id }}\nPrimary: ${{ env.PRIMARY_REGION }} | DR: ${{ env.DR_REGION }}\nRTO target: 900s | RPO target: 300s\nView details: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}"
        env:
          SLACK_WEBHOOK_URL: ${{ secrets.SLACK_WEBHOOK_URL }}

      - name: Notify on success
        if: steps.dr-test.outcome == 'success'
        uses: slackapi/slack-github-action@v1
        with:
          slack-message: "✓ *DR Test Passed* — ${{ github.repository }}/${{ github.run_id }}\nPrimary: ${{ env.PRIMARY_REGION }} | DR: ${{ env.DR_REGION }}\nAll RTO/RPO targets met."
        env:
          SLACK_WEBHOOK_URL: ${{ secrets.SLACK_WEBHOOK_URL }}

      - name: Create GitHub issue on failure
        if: steps.dr-test.outcome == 'failure'
        uses: actions/github-script@v7
        with:
          script: |
            github.rest.issues.create({
              owner: context.repo.owner,
              repo: context.repo.repo,
              title: `DR Test Failed — ${context.runId}`,
              body: `## Disaster Recovery Test Failure\n\n**Run**: ${context.serverUrl}/${context.repo.owner}/${context.repo.repo}/actions/runs/${context.runId}\n**Primary Region**: ${process.env.PRIMARY_REGION}\n**DR Region**: ${process.env.DR_REGION}\n**Date**: ${new Date().toISOString()}\n\nThe monthly DR test has failed. Review the test report and take corrective action.`,
              labels: ['dr-test', 'critical']
            })