.gitlab-ci.yml

# inspired from https://docs.gitlab.com/ee/user/packages/container_registry/
# define the stages for this CI as well as the images to pull and services to use
image: docker:20.10.21
services:
  - docker:20.10.16-dind
stages:
  - build
  - release
  - gke-deploy

# declare the variables used within the CI and eventually docker files
variables:
  WORKDIR: "news_bias_evaluator"
  IMAGE_TAG: $CI_REGISTRY_IMAGE:$CI_COMMIT_REF_SLUG # name used to build the docker image
  LATEST_RELEASE_IMAGE_TAG: $CI_REGISTRY_IMAGE:latest # name of the docker image using latest label
  VERSION_RELEASE_IMAGE_TAG: $CI_REGISTRY_IMAGE:$CI_COMMIT_TAG # name of the docker image using the commit tag label (for versioning and treacability)
  KUBE_CONTEXT: "courses/dit825/2022/group07/group-07:dit825-cluster-agent" # location of the k8s cluster agent that will be used for deploying to gke.

# setup the environment before running the different scripts
default: #default lets us exclude the before script from the gke-deployment stage
  before_script:
    # update an upgrade
    - apk update
    - apk upgrade
    # install docker compose
    - apk add docker-compose
    # Print info to check that everything is installed as it should
    - docker --version
    - docker-compose --version
    # login to the docker deamon
    - echo $CI_REGISTRY_PASSWORD | docker login -u $CI_REGISTRY_USER --password-stdin $CI_REGISTRY
    # go into working directory
    - cd $WORKDIR

# run the build step
build:
  # define which runner to use
  tags:
    - docker-build
  stage: build
  script:
    #build the docker image using docker compose
    - docker-compose build

#run the release step
release:
  # define which runner to use
  tags:
    - docker-build
  stage: release
  rules:
    # add a rule to only run if there is a tag created
    # and if the tag follows the given regex (example format v10.12.06)
    - if: $CI_COMMIT_TAG =~ /^v\d+\.\d+\.\d+/
  script:
    # build the docker image using docker compose
    - docker-compose build
    # add the tag "latest" to the image for release
    - docker tag $IMAGE_TAG $LATEST_RELEASE_IMAGE_TAG
    # push to registry
    - docker push $LATEST_RELEASE_IMAGE_TAG
    # Use the commit tag to save separate tag version (both latest and version tag can be used when pulling the image)
    - docker tag $IMAGE_TAG $VERSION_RELEASE_IMAGE_TAG
    # push to registry
    - docker push $VERSION_RELEASE_IMAGE_TAG

deploy:
  # gke-deployment job wont inherit any default: keys (in this case, the before_script)
  inherit:
    default: false
  stage: gke-deploy
  # Run job only when the release finishes
  needs: [release]
  # Define an image in which the kubectl cli can be run
  image:
    name: bitnami/kubectl:latest
    # Override the default entry point of the bitnami/kubectl image to
    # allow the runner to start the container in a desired way.
    # more info: https://gitlab.com/gitlab-org/gitlab-foss/-/issues/65110
    entrypoint: [""]
  # define which runner to use
  tags:
    - docker-build
  script:
    # get information about the current k8s contexts.
    # NOTE: a context groups access parameters to a k8s cluster.
    - kubectl config get-contexts
    # switch to the K8s context provided by our gitlab gke agent - this allows us to
    # access the k8s cluster on GKE.
    - kubectl config use-context "$KUBE_CONTEXT"
    # Apply the defined K8S manifest.
    # NOTE: This manifest defines the deployment of the django app, and the load
    # balancer service (which enables external access).
    - kubectl apply -f $CI_PROJECT_DIR/k8s_manifests/news-bias-evaluator-app.yaml
  rules:
    # same reasoning as the release stage.
    # run the gke-deploy stage only if a tag is
    # created with the regex pattern
    # The same rule is needed as the release stage in order for the "needs: [release]" line to work.
    # more info: https://gitlab.com/gitlab-org/gitlab/-/issues/370052
    - if: $CI_COMMIT_TAG =~ /^v\d+\.\d+\.\d+/