Canary releases with progressive traffic shifting

Deploying directly to 100% of traffic means a bad release hits all users immediately. By the time monitoring alerts fire, thousands of users have already seen the error.

Route a small percentage of traffic to the new version and increase gradually based on error rate:

jobs:
  canary-deploy:
    runs-on: ubuntu-latest
    steps:
      - name: Deploy canary (5% traffic)
        run: |
          kubectl apply -f manifests/canary.yaml
          kubectl patch virtualservice myapp --type merge -p \
            '{"spec":{"http":[{"route":[{"destination":{"subset":"v2"},"weight":5},{"destination":{"subset":"v1"},"weight":95}]}]}}'

      - name: Monitor error rate for 5 minutes
        run: |
          sleep 300
          ERROR_RATE=$(./get-error-rate.sh canary)
          if (( $(echo "$ERROR_RATE > 1.0" | bc -l) )); then
            echo "Error rate $ERROR_RATE% too high, rolling back"
            ./rollback-canary.sh
            exit 1
          fi

      - name: Promote to 100%
        run: ./promote-canary.sh

5% canary limits blast radius to 5% of users. If error rates spike, rollback is immediate. Automated monitoring removes the need for a human to watch dashboards during every deploy.