Skip to main content
In this guide, we’ll walk through adding a dev-deploy stage to your existing GitLab CI pipeline. This stage installs the kubectl CLI on an Alpine runner and deploys Kubernetes manifest files to your cluster.

​
1. Define Pipeline Stages

First, extend your .gitlab-ci.yml to include the new deploy stage alongside test and containerization: 
stages:
  - test
  - containerization
  - dev-deploy

variables:
  DOCKER_USERNAME: siddharth67
  IMAGE_VERSION: $CI_PIPELINE_ID
StagePurposeExample Job
testRun unit tests and code coverageunit_tests
containerizationBuild and push Docker imagesbuild_image
dev-deployInstall kubectl and deploy manifests to Kubernetesk8s_dev_deploy
Make sure the DOCKER_USERNAME and IMAGE_VERSION variables align with your project settings.

​
2. Create the k8s_dev_deploy Job

Add a job in the dev-deploy stage that uses Alpine 3.7, installs kubectl, and avoids pulling artifacts from earlier stages: 
k8s_dev_deploy:
  stage: dev-deploy
  image:
    name: alpine:3.7
  dependencies: []
  before_script:
    # Download and install the latest stable kubectl binary
    - wget -qO- "https://storage.googleapis.com/kubernetes-release/release/$(wget -qO- https://storage.googleapis.com/kubernetes-release/release/stable.txt)/bin/linux/amd64/kubectl" -O kubectl
    - chmod +x kubectl
    - mv kubectl /usr/bin/kubectl
  script:
    - kubectl version -o yaml

​
3. Visualize the Pipeline

Once committed, your pipeline will include three stages—test, containerization, and dev-deploy. The diagram below shows how the new dev-deploy stage sits at the end of the workflow:
The image shows a GitLab Pipeline Editor interface with a visual representation of a CI/CD pipeline, including stages like testing, containerization, and deployment. The pipeline includes steps such as unit testing, code coverage, docker build, and Kubernetes deployment.

​
Speed Up Iterations

Since this deployment job doesn’t depend on artifacts from earlier stages, you can temporarily comment out other jobs (e.g., using Ctrl+/). This lets you focus solely on dev-deploy during development.
The image shows a GitLab CI/CD pipeline interface for a project named "Solar System NodeJS Pipeline," indicating a failed job in the "dev-deploy" stage. The sidebar displays various project management options like issues, merge requests, and pipelines.

​
4. Troubleshoot the kubectl version Error

In the job logs, you’ll see the kubectl binary download successfully, but the version command fails to retrieve the server version: 
$ wget -qO- https://storage.googleapis.com/kubernetes-release/release/$(wget -qO- https://storage.googleapis.com/kubernetes-release/release/stable.txt)/bin/linux/amd64/kubectl -O kubectl
$ chmod +x kubectl
$ mv kubectl /usr/bin/kubectl
$ kubectl version -o yaml
clientVersion:
  buildDate: "2024-01-17T15:51:03Z"
  compiler: gc
  gitCommit: bc40b19f2782041b3fb39fcaf3a995c4de90d2
  gitTreeState: Clean
  gitVersion: v1.29.1
  major: "1"
  minor: "29"
  platform: linux/amd64
kustomizeVersion: v5.4.0-2023601165947-6ce0bf390ce3
ERROR: Job failed: exit code 1
This error indicates that kubectl cannot connect to a Kubernetes API server without a valid kubeconfig file. You must configure the kubeconfig before running any cluster operations.

​
5. Next Steps

  1. Add your cluster credentials to a GitLab CI/CD variable (e.g., KUBE_CONFIG).
  2. Write the kubeconfig file in a before_script step.
  3. Re-run the pipeline to confirm that kubectl version returns both client and server information.

​
References

Watch Video