Customizing the Build Environment

Semaphore runs each job in a clean and isolated environment. The default Ubuntu environment includes common tools needed by most projects. With unrestricted sudo access and fast network, you can install additional software and customize the environment however you need.

Installing additional dependencies

Let's say that your project uses Bats, which is hosted on GitHub, to test your Bash scripts. You can easily make it available on Semaphore by copying the installation commands from the project's Readme into your prologue:

# .semaphore/semaphore.yml
blocks:
  - name: "Test"
    task:
      prologue:
        commands:
          - git clone https://github.com/bats-core/bats-core.git
          - cd bats
          - sudo ./install.sh /usr/local
      jobs:
        - name: Tests
          commands:
            - bats addition.bats

Using databases and other services

Let's say that your CI build needs Redis and PostgreSQL:

# .semaphore/semaphore.yml
blocks:
  - name: "Test"
    task:
      prologue:
        commands:
          - sem-service start redis
          - sem-service start postgres
      jobs:
        - name: Tests
          commands:
            - createdb -U postgres -h 0.0.0.0 myapp_database
            - echo 'running tests'

To manage database engines and other services on Semaphore, use the sem-service utility.

Since you have unrestricted access to the job's environment, other options for running services include installing native packages with sudo and using Docker Compose.

Using services and test data across blocks

Note that, since all jobs run in isolated environments, the services that you start in one block are not automatically available in other blocks. Likewise, starting a service in one job, doesn't automatically make it available in other jobs of the same block.

To use a service or populate test data in all parallel jobs within a block, specify that in the task prologue. Repeat the same steps in the definition of each block as needed.

Next steps

Almost every project has dependencies, and we can save a lot of time by installing them once and reusing them from a cache. Let's learn how to do that in the next section.

Still need help? Contact Us Contact Us