Technology
CI/CD
Continuous integration and delivery: automated tests, builds, safe releases and fast rollback.
About the technology
I have been building CI/CD for seven years, because I am convinced that the speed and safety of delivering changes determine the health of the whole engineering process. Continuous integration in my projects means that every commit automatically goes through a build and a full set of checks: static analysis (PHPStan, golangci-lint), code-style checks (Pint, gofmt), unit and integration tests, and a dependency audit for vulnerabilities. This way problems are found in minutes, not in production, and the branch always stays in a working state. I set up pipelines in GitHub Actions and GitLab CI: I split them into parallel stages, cache dependencies and Docker layers to speed things up, and use matrices to test against different environment versions. I move artifact building into a separate stage — I build and tag Docker images and publish them to a registry with meaningful versions tied to the commit. I organize continuous delivery with environment separation: automatic deployment to staging for testing and manual or automatic promotion to production through protected environments with mandatory review. I make releases safe: I apply rolling-update, canary and blue-green strategies, run database migrations in a controlled way with backward compatibility in mind, warm up configuration and route caches, and check service health with automated smoke tests right after rollout. The ability to roll back quickly is a mandatory requirement for me: versioned images and infrastructure as code let me return to the previous state in seconds. I keep secrets out of the repository, pass them through secure CI vaults and secret managers, and never log or commit them. The GitOps principle, where the desired infrastructure state is described in Git and applied automatically, gives reproducibility, audit and transparency of changes. I track the delivery process own metrics — release frequency, time from commit to production, the share of failed deployments and recovery time — and continuously shorten the feedback loop. A well-configured CI/CD in my practice turns a release from a stressful event into a routine, predictable and reversible operation, which directly speeds up the delivery of value to the business.
Experience
Projects using this technology
There are no published projects with this technology yet. See all projects.