Build a Full CI/CD Pipeline for a Sample App
Part of the DevOps Roadmap
In short
Time to wire everything from this phase into one pipeline you can show in interviews: a push to main tests, builds, stores and deploys your app with nobody touching a server.
The practice project for this phase — take a small sample application from commit through automated build, test, and deployment, wiring the full pipeline together end to end.
Think of it like this
Tracking a food delivery order
Your delivery app shows each step: order placed, restaurant cooking, picked up, on the way, delivered. If it stalls, you see exactly where. Your pipeline should feel the same: every stage visible, each one gated on the last, and you know where it stopped.
Watch it happen
Your build plan, stage by stage
1/6Every push and PR runs lint and unit tests. A failure stops everything after it.
The big ideas
What you need to know
Tests gate everything
No image gets built unless tests pass. That's the safety net the rest of the pipeline relies on.
Immutable tags
Tag images with the commit SHA and deploy that same tag everywhere. No latest.
Secrets done right
Registry and server credentials live in the secret store or come from OIDC. Nothing in the YAML.
A rollback plan
Write down how to redeploy the previous tag. Even better, make it one pipeline run.
See it in code
Walk through the code
1name: pipeline2on:3 push:4 branches: [main]5jobs:6 test:7 runs-on: ubuntu-latest8 steps:9 - uses: actions/checkout@v410 - run: npm ci && npm test11 image:12 needs: test13 runs-on: ubuntu-latest14 steps:15 - uses: actions/checkout@v416 - run: docker build -t ghcr.io/your-name/sample-app:${{ github.sha }} .1/4Run the whole pipeline on every push to main.
This is the skeleton. Next, add a registry login and push step, then staging and production deploy jobs that each need the job before them.
Save yourself the pain
Common mistakes
Oops
Building one giant job that does test, build and deploy in a single script.
Do this instead
Split into jobs with needs:. You see exactly which stage failed and can rerun just that one.
Oops
Deploying to production straight from every push with no check after deploy.
Do this instead
Add a staging step, a smoke test and an approval gate before production. Then check /health after deploying.
Your mission
Milestone 1: green to staging
Quick check
In GitHub Actions, what does needs: test on the image job do?
Go deeper
- 1Choose a small sample app (even a 'hello world' service) to keep focus on the pipeline itself.
- 2Include build, automated tests, and a deployment step to somewhere real (even a free-tier host).
- 3Document your pipeline's stages in a README so someone else could understand it at a glance.
Stuck on this topic? Ask an Insider
Get 1:1 guidance from people who've walked this exact path — on the InsideEdge app.
