crudeci

crudeci is a rudimentary CI/CD system written in Go.

There are three main constructs in crude

• Job: represents a typical job in a CI/CD system
• Pipeline: represents a sequence of steps to be executed as part of a job.
• Step: represents a single step of execution in a pipeline.

The ASCII diagram below visually represents a single job construct along with its sub-components:

+------------------------------------------------------------------+
| sample-job                                                       |
| |                                                                |
|  \-- Pipeline:                                                   |
|      +---------------------------------------------------+       |
|      | Steps:                                            |       |
|      |    +-------+     +-------+              +-------+ |       |
|      |    | build |     | test  |    ....      | step k| |       |
|      |    +-------+     +-------+              +-------+ |       |
|      +---------------------------------------------------+       |
|                                                                  |
+------------------------------------------------------------------+

Walk-through

Upon starting, crudeci reads and parses the default configuration file, config.json, where jobs are declared and configured. crudeci provides a -config flag for the user to specify a non-default configuration file if desired.

With the corresponding structs populated, crudeci (1) initializes each of the jobs by cloning the job repository if it does not already exist and (2) executes each of the steps declared in the pipeline within embedded in each of the jobs. See the Job struct in job.go. All assets resulting from the above two steps are placed in the out/ sub-directory by default but user allowed to specify via the -outdir flag.

The job pipelines in this project are rather simple, consisting of (1) build and (2) test steps, making this codebase more of a CI system than a CD system. That being said, the pipelines and steps are extensible. For example, to implement a deploy step - in step.go - declare a DeployStep struct, and satisfy the Runner interface by implementing the Run() method accordingly.

Configuration

Below is a snippet from config.json:

[
    {
        "name": "sample-job",                        # Job name
        "repo": "https://github.com/golang/example", # Job repo
        "workingDir": "outyet/",                     # Job working directory
        "build": "go build main.go",                 # Build step
        "test": "go test -v"                         # Test step
        "notify": "[email protected]",           # Job notification contact email
    }
]

Requirements

• Docker

Running

From the top level of this repository, run:

docker build -t crudeci .                      # to build Docker image
docker run -it --rm crudeci sh -c "make run"   # to run Docker container

Limitations

• Program executes in sequential manner for simplicity, though rudimentary concurrent approach is provided in the runConcurrently() function (commented out at the end of main.go).
  • Thus, program terminates and thus does not watch for code changes to repositories.
    • Extending the Step struct by creating a PullStep struct and satisfying the Runner interface is an acceptable approach.
• Job cleanup functionality missing but can be implemented by creating a CleanupStep and satisfying the Runner interface with the appropriate logic.
• User not messaged with job status. As with limitations above, functionality can be implemented by extending the Step struct.
• Program doesn’t preserve state.
• Program output not stored for bookkeeping. Currently logged to the terminal.
• Program output formatting has room for improvements.
• The structures declared are not in the optimal format.