ramizouari / Excellentia

A Platform for sharing computer science knowledge and improving problem solving

Excellentia

A Platform for sharing computer science knowledge and improving problem solving.

1. Introduction

Excellentia is a platform for exploring computer science knowledge, and it also hosts some probmes that help individuals to sharp their skills.

Created as a academic project, as of now, it only has backend due to time pressures.

2. Back-end

The back-end consists of:

1. A Compiler, that potentially compiles a solution
2. A Runner, that runs a proposed solution for a problem on some test cases and issues a Verdict (Accepted, Wrong Answer)
3. A Judge, which acts as an orchestrator, it calls the compiler to compile the proposed solution's code, and then calls the runner to run the solution and check its correctness

The judge will act as the main entry point for the project.

For now, the hosted problems are The WinterCup 4.0 official problems, which was hosted on 09 April 2022, at INSAT.

The back-end is implemented using Spring Boot 3.0, and supports:

• Logging
• Metrics using Prometheus
• Health status
• Traces (Experimental)

3. Provisioning

3.1 Database

We used a MS-SQL instance as a database, it is used to store:

• Problems: the information about problems
• Runs: the information about a proposed solution

Its deployment is described at

3.2 Excellentia Server

We used an Azure Kubernetes Service instance to host the application. Its provisioning described here

3.3 Storage

To ease the deployment, we hosted a Blob Storage on which we uploaded a 7-zip file containing the test cases of all Winter Cup problems.

The name of the file is tests.7z, and it is located at the tests container.

4. Deployment

Once the provisioning is successfully completed, the application will be deployed on 2 main phases:

4.1 Dockerization

On this phase, we created 4 docker images:

• excellentia-common which acts as a common image, It has common configurations
• excellentia-judge which is the image containing the Judge micro-service
• excellentia-compiler which is the image containing the Compiler micro-service
• excellentia-runner which is the image containing the Runner micro-service

To ease dockerization, we implemented a helper bash script at backend sub-directory that creates all images

./dockerize-all.sh

Once dockerized, we implemented a helper bash script at backend sub-directory that deploy all images to Docker Hub

./push-all-docker.sh ACCOUNT_NAME

4.2 Kubernetes Deployment

On this phase, we created a Helm chart, which is located at deployment/4-kubernetes. It contains:

• The required instances, with:
  • $2$ pods for Judge
  • $2$ pods for Compiler
  • $4$ pods for Runner
  • $1$ Ingress pod
  • $1$ Prometheus pod
• The shared disks (Exactly, the Persistent volumes claims):
  • The disk RUNNER_DISK on which runs are stored
  • The disk TESTS_DISK on which tests are stored
  • The disk LOGS_DISK on which logs are stored
• An InitContainer, that runs on the compiler pod that:
  • Installs curl and p7zip-full
  • Downloads the tests.7z file located at the Blob Storage
  • Extracts it
  • Copy it to TESTS_DISK
• A Prometheus instance for collecting metrics
• An Ingress controller acting as a gateway

5. API

For now, we proposed the following API for consumption:

• /judge/status to get the status of the judge
• /judge/{problem_id} to submit a solution for the problem identified byproblem_id. It returns a solution verdict
• /actuator/health, to check the health of the micro-service
• /actuator/prometheus, to collect the metrics of the micro-service
• /actuator/logsfile, to collect the logs of the micro-service

6. Architecture