A Lightweight Mini Supercomputer Testbed
PiHPC is a barebones, lightweight high-performance computing (HPC) cluster testbed built for experimentation, demos, and hands-on learning.
This repository serves as a launchpad for HPC systems design using modern infrastructure tooling such as Ansible, Kubernetes, SLURM, Splunk, and Grafana.
Think of it as a mini supercomputer lab you can run anywhere — perfect for testing configs, workflows, and monitoring pipelines before scaling to real HPC environments.
O.o
- Infrastructure-as-Code with Ansible for cluster provisioning
- Containerized workflows via Kubernetes
- ⏱Job scheduling with SLURM
- Observability stack with Grafana, Prometheus, and Splunk
- Scalable design demos — networking configs, storage backends, and cluster layout examples
- Pluggable examples: toy workloads, monitoring dashboards, and cluster workflows
pihpc/
├── ansible/ # Cluster config + provisioning playbooks
├── kubernetes/ # Manifests, Helm charts, containerized HPC demos
├── slurm/ # SLURM configs, sample job scripts
├── monitoring/ # Grafana dashboards, Prometheus, Splunk queries
├── demos/ # Lightweight HPC-style test apps and workflows
└── README.md- Linux (Ubuntu/Debian/Rocky recommended)
- Python 3.x
- Ansible
- Kubernetes or Minikube
- SLURM
# Clone the repository
git clone https://github.com/<your-username>/pihpc.git
cd pihpc
# Run base Ansible provisioning
ansible-playbook ansible/cluster.yml
# Deploy Kubernetes components
kubectl apply -f kubernetes/
# Launch monitoring stack
kubectl apply -f monitoring/- Submit a SLURM job to run a toy MPI benchmark
- Monitor cluster health & metrics in Grafana
- Send logs and job events into Splunk for indexing & searching
- Deploy a containerized workload with Kubernetes and compare performance vs. bare metal
- Ansible – Automation & configuration management
- Kubernetes – Container orchestration
- SLURM – HPC workload manager
- Grafana – Dashboards & visualization
- Prometheus – Metrics collection & alerting
- Splunk – Logging & observability
- Provide a sandbox environment for HPC concepts
- Explore modern DevOps + HPC workflows (containers, observability, IaC)
- Share config examples for reproducible, lightweight clusters
- Serve as a learning resource for HPC engineers and enthusiasts
- Contributions, PRs, and suggestions are welcome!
- Got a new demo workload, monitoring integration, or optimization tip? Open an issue or send a pull request.
MIT License – Free to use, modify, and share.
PiHPC — because sometimes you just need a tiny supercomputer to play with.