Runner Engine

A minimal, high-performance continuous integration (CI) pipeline execution engine written in Go.

The platform securely evaluates multi-stage workflow pipelines parsed from graph architectures concurrently using an isolated containerized infrastructure sandbox.

Core Architecture & Design (v1)

The platform separates the orchestration layer from the execution environment to ensure secure, parallel, and predictable multi-stage workflow execution

Why gVisor + Docker?

Standard container architectures share the host system's Linux kernel directly. If a developer runs a malicious pipeline script or hits an unknown dependency flaw, they can escape the container space and corrupt the core server.

• gVisor Integration: This platform wraps container workloads inside a custom runsc application kernel virtualization layer. It intercepts unprivileged host system calls, completely shielding the host kernel from potential multi-tenant execution panics or security compromises.
• Docker Context: Docker is used to manage dynamic image layer replication (python, alpine, etc.) and provide an ephemeral volume lifecycle, mounting active workspaces directly onto isolated paths.

Getting Started

Prerequisites

• Go: 1.22+
• Docker Engine (with gVisor runtime configured)

Local Configuration Setup

1. Clone the Repository:

   git clone https://github.com/drona-gyawali/runner.git
   cd runner
   

2. Compile the Core Binary:

   go build -o bin/runner cmd/runner/main.go
   

3. Execute a Workflow Pipeline:

To run a pipeline workflow, pass the target configuration file path using the -config flag:

    ./bin/runner -config YOUR_FILE_PATH/runner/workflows/YOUR_CI_NAME.toml

Integration Test

The engine utilizes a zero-cache integration test harness that tests the compiled binary against live project repositories under realistic runtime conditions.

The test suite runs an end-to-end project fixture validation matrix. It spins up concurrent jobs and forces a dependent bottleneck sync, and evaluates core unit tests.

Running the Integration Suite

Execute the following command in your terminal space:

go test -v ./tests/...

This engine was engineered completely from scratch out of a deep first-principles curiosity to understand the inner workings of distributed infrastructure, Docker internals, and sandbox virtualization layers.

If you appreciate the architecture or find the implementation patterns helpful, drop a star on the repository! It helps keep the project visible to other systems engineers. ⭐