How to Run a Smoke Test After Deploy in GitHub Actions

A smoke-test job that needs: the deploy polls the live endpoint and fails the run if the release is bad.

Add a job that needs: deploy and curls a health or canary endpoint with retries. If it never returns healthy, the job fails - your signal to roll back.

Poll health after deploy

The smoke job runs only after deploy; the retry loop fails the job if health never comes up.

jobs:
  deploy:
    runs-on: ubuntu-latest
    steps:
      - run: ./deploy.sh
  smoke:
    needs: deploy
    runs-on: ubuntu-latest
    steps:
      - run: |
          for i in $(seq 1 10); do
            code=$(curl -s -o /dev/null -w '%{http_code}' https://app.example.com/healthz)
            [ "$code" = "200" ] && exit 0
            sleep 6
          done
          echo "Health check never returned 200" && exit 1

Gotchas

• Give the new release time to come up - poll with backoff rather than a single immediate request.
• Hit a real readiness signal (/healthz, a canary route), not just the load balancer, which may answer before pods are ready.
• Pair the smoke test with an automatic rollback step (if: failure()) so a bad release does not stay live.

Related guides