Continuous integration, delivery, and deployment (CI/CD)

Although CI/CD is commonly used for various types of testing in industry, I prefer not to overly rely on it to run my feedback scripts. One reason is that I want to be able to conveniently run the scripts manually on my local machine.

Therefore, GitLab’s CI/CD serves only two purposes:

• To invoke a sandbox environment, such as a GitLab Runner Docker container, to run the feedback scripts
• To provide a single entry point (e.g., feedback.sh) for the feedback process. The container runs ./feedback.sh and then collects the results.

The logic for generating feedback is entirely scripted within the ece100/root/a0/assessment repository.

GitLab Job Runners

A GitLab Runner is an executor that runs CI/CD jobs on machines outside of GitLab. There are two main steps to create a runner:

1. Register the runner on GitLab This step involves creating a runner entry in GitLab. It’s essentially requesting a runner token or ID that the actual runner will use to authenticate and communicate with GitLab.

2. Set up the actual runner on an external machine The runner can be installed on your laptop or any other machine. For simplicity, I will create a GitLab Runner Docker container running on my laptop.

See the execution flow

Create a runner on GitLab

Here I create a shared runner for the entire course, the group_id is the same as above.

I set run_untagged to true for simplicity. It means the runner can be used for any repo inside the group.

glab api --method POST /user/runners \
    --field runner_type="group_type" \
    --field group_id=94733 \
    --field description="ece100_runner_0" \
    --field run_untagged=true

# output
{
  "id": 1245,
  "token": "*************************",
  "token_expires_at": null
}

The runner authentication token will only show up here, so make sure you note it down.

In case you forgot to note it down, you can reset it by

glab api --method POST /runners/1245/reset_authentication_token

Create the gitlab-runner docker container

With the docker engine running on my local machine, then I do:

TOKEN=<RUNNER_TOKEN>

docker volume create gitlab-runner-config

# Step 1: start a gitlab-runner daemon (run in a container)
docker run -d --name gitlab-runner --restart always \
    -v /var/run/docker.sock:/var/run/docker.sock \
    -v gitlab-runner-config:/etc/gitlab-runner \
    --env TZ=America/Toronto \
    gitlab/gitlab-runner:latest

# Step 2: register runner workers
# Given the executor is "docker", so the job is executed in
# `dind` (Docker-in-Docker) mode
docker run --rm -it \
    -v gitlab-runner-config:/etc/gitlab-runner \
    gitlab/gitlab-runner register \
    --non-interactive \
    --url "https://git.uwaterloo.ca" \
    --token "$TOKEN" \
    --executor "docker" \
    --docker-image ubuntu:latest

If gitlab-runner is in a container while you still want to use docker as the executor, then you need to provide the login info for gitlab-runner to access the docker hub.