Industrialising the delivery pipeline and moving to GitOps

Context

Seven Java applications and three front-end applications delivered by one IT department, with different integration pipelines per team. Releases happened on Friday evenings, with a handwritten deployment file and manual server interventions.

Problem

A release required six hours of work, failed one time in four, and rollback relied on restoring a database backup. No reliable record showed which version was deployed in which environment.

Objectives

• Standardise integration and delivery pipelines across teams
• Make deployments reproducible and reversible
• Shorten the delay between a validated fix and its release

Role

Owner of the industrialisation workstream, with occasional support from an external security architect on the dependency control chain.

Solution

The initial diagnosis measured the existing pipelines: an average of seventeen minutes for tests, ninety-two minutes for the full pipeline, with a twenty-three percent failure rate of which a third came from infrastructure causes. Those measurements became the baseline throughout the project. A shared pipeline was built from reusable actions: compilation, parallel unit tests, blocking static analysis on sensitive points, container-based integration tests, and building and publishing signed images. End-to-end tests were repositioned to critical journeys so as not to lengthen the validation path. Deployment moved to GitOps: manifests live in a dedicated repository, Argo CD applies the described state and detects any drift. The handwritten deployment file was replaced by a versioned procedure, and rollback means reverting to the previous version of the configuration repository.

Architecture

GitHub Actions pipelines for applications, with reusable actions shared between teams and managed dependency caching. Images are built in multi-stage mode, scanned and signed, then stored in an internal registry. The Kubernetes cluster is described with Terraform, and its application content by versioned manifests applied by Argo CD, with one configuration repository per environment. Security checks run on every delivery: dependencies, base image, accidentally committed secrets and static analysis.

Results (samples)

• Shared integration pipeline across ten applications
• Versioned deployments with rollback by reverting to the previous version
• Security checks running automatically on every delivery
• Written release procedure applied without manual intervention

Indicators (samples)

• Average full pipeline duration

  92 min to 21 min

  Simulated measurements — demonstration sample

• Release failure rate

  1 in 4 to 1 in 12

  Indicative estimate on a simulated scope — demonstration sample

• Releases delivered over the period

  22

  Illustrative volume — demonstration sample

Technologies

• Docker
• Kubernetes
• Terraform
• GitHub Actions
• GitLab CI
• Argo CD
• JUnit 5
• Playwright
• SonarQube

Related services

• DevOps and CI/CD delivery pipelines

  I build delivery pipelines that genuinely test the code and allow rollback within minutes. Deployment stops being an event and becomes a routine operation.

• Docker containerisation and Kubernetes orchestration

  I containerise your applications with lean images and run them on Kubernetes with controlled manifests. The platform becomes predictable, properly sized and repairable.

• Application security and API protection

  I secure your applications where it actually matters: identities, authorisations, secrets and input validation. The measures are verifiable and integrated into the delivery pipeline.

• Training and coaching for technical teams

  I train your teams on your own projects, with exercises drawn from your code. Sessions end with usable output, not a satisfaction survey.