Using lock file as Gitlab CI/CD cache key

Gitlab CI/CD provides a mechanism to cache project dependency so that it can be reused and shared between pipelines and jobs. Using cache decreases the time spent on each job. The cache can be enabled by using cache: keyword in your .gitlab-ci.yml file. It also allows for providing a cache:key directive for each cache. Different strategies can be used for generating cache keys.

e.g: Using predefined variables like $CI_COMMIT_REF_SLUG to create separate cache for each branch.

But projects using npm have their lock file for managing dependency which can be used to create a cache key. The cache will only be updated when the lock file is changed. By adding cache:key directive the key is saved as the SHA checksum of the lock file, therefore key will only change when the lock file is updated. Even if there are many branches in your project unless the packages are updated or new one are added they can all share the same cache.

Here is the example of using package-lock.json as a cache key which is generated automatically when you install any packages.

stages:
  - ci

.cache: &cache
  cache:
    key:
      files:
        - package-lock.json
    paths:
      - node_modules

lint:
  stage: ci
  image: node:12.14.1
  <<: *cache
  script:
    - npm install
    - npm run lint

test:
  stage: ci
  image: node:12.14.1
  <<: *cache
  script:
    - npm install
    - npm run test

If your project uses yarn this you can change to use yarn.lock as key

stages:
  - ci

.cache: &cache
  cache:
    key:
      files:
        - yarn.lock
    paths:
      - node_modules

lint:
  stage: ci
  image: node:12.14.1
  <<: *cache
  script:
    - yarn
    - yarn lint

test:
  stage: ci
  image: node:12.14.1
  <<: *cache
  script:
    - yarn
    - yarn test

Now gitlab will save and reuse the content of the cache on each subsequent jobs.

Reference:

• Gitlab CI/CD Cache Key