Bringing Old Giants Back to Life: The Ranger Cluster Story at Sol Plaatje University

Bringing Old Giants Back to Life: The Ranger Cluster Story at Sol Plaatje University

July 06, 2026   Post, Publication,

The ranger cluster racks at Sol Plaatje University, once dormant, now reborn for modern research  

What happens when you hand a determined team a room full of dormant, two-decade-old servers? At Sol Plaatje University (SPU), you get Ranger, a high-performance computing (HPC) cluster that proves innovation doesn’t always mean buying new.

Led by Gift Nemakonde and guided by Dr Silas Verkijika, this project is all about making the most of what you have. Instead of sending ageing Sun Blade hardware to the scrapyard, the SPU team rolled up their sleeves and got creative, breathing new life into forgotten machines with a modern software stack and a whole lot of grit.

Cluster By the Numbers

• Nodes Online: 41 out of 48 possible (system can scale to 96)

• CPU Power: 656 AMD Opteron cores (16 cores/node)

• Memory: 1.3 TB RAM (32 GB/node)

• Storage: 11 TB shared NFS storage

• Headnode: Dell PowerEdge T340

• Chassis: Sun Blade X6048 (2 racks, 8 chassis)

From Dusty to Dynamic

It wasn’t just a matter of plugging things in and flipping a switch. Each step was a battle against time, technical quirks, and the realities of legacy systems. The team rebuilt kernel drivers to restore dropped network connectivity, re-engineered the PXE/iPXE boot process to avoid RAM bottlenecks, and brought 41 of the 48 blades online, while work continues to bring the full system to stable production capacity. Instead of decompressing large OS images in memory, nodes now mount minimal images via NFS, slashing boot-time memory requirements and avoiding out-of-memory panics.

Modern Tech, Classic Machines

Ranger runs Rocky Linux 8, uses OpenHPC 2.x, and is managed with Warewulf 4 provisioning and Slurm as its scheduler. The cluster delivers all user software via centralised and stateless NFS shares, including advanced tools like GCC 12, OpenMPI 4.1.6, OpenBLAS, and Miniconda3 Python 3.12. Every detail is tuned for both reliability and efficiency.

Teamwork, Tenacity, and Ingenuity

Challenges weren’t just technical; they were logistical. Each blade required manual boot order configuration. Standardising Munge keys and enforcing NTP clock sync ensured node authentication and uptime. Moving to nolock NFS mounts eliminated frustrating mount failures and job script errors. Every problem was met with teamwork and creativity, turning setbacks into innovative solutions.

Performance That Defies the Calendar

Despite hardware dating back to 2008, Ranger punches far above its weight:

• Monte Carlo Pi (1 billion iterations): 40x speedup across 41 nodes, maintaining 97.9% parallel efficiency

• Matrix Multiplication (10,000 x 10,000): Completed reliably with no node failures

• Network Latency: 38.6 microseconds (exceptional for 1 GbE)

• Aggregate Write Throughput: 140 MB/s under parallel load

Ranger is now entering the stage of onboarding research projects across disciplines.

A Lesson in Making It Work

Ranger is more than just racks and cables. It’s proof that with enough skill and heart, yesterday’s technology can fuel today’s discoveries. At SPU, the Ranger cluster is a celebration of ingenuity, teamwork, and the belief that the best solutions often start with what you already have. As members of the HPC Ecosystems Project, we are proud to see these resources revitalised and delivering real impact for research and innovation at SPU.

#HPC #HighPerformanceComputing #Cluster #SPU #CHPC #OpenHPC #Linux #HPCecosystems #Innovation #Research

Continue Reading

CHPC Empowers Wits University with Collaborative HPC Equipment Allocation

CHPC Empowers Wits University with Collaborative HPC Equipment Allocation

April 07, 2026   Post,

New Infrastructure for Wits Research: Excitement is buzzing at the University of the Witwatersrand (Wits) as a major boost to their computing capacity arrives on campus. As part of the HPC Ecosystems Project under NICIS, selected HPC computing equipment has been transferred to Wits for use within the Student Interest Group in High‑Performance Computing (SIG HPC). This initiative aims to strengthen HPC skills development, practical training, and student-led technical capacity building. Wits researchers are receiving a comprehensive suite of 34 high-density computing components, including a Dell PowerEdge M1000e chassis and a fleet of M600, M610, and M805 blade servers.

The Logistics of Success: Moving a full-scale server infrastructure deployment is a precision operation. Behind the seamless transfer was our dedicated colleague, Samuel Mathekga, who managed the process locally, ensuring all equipment was delivered and set up according to plan

We also want to recognise the CHPC logistics team, whose expertise in handling sensitive, high-value hardware was crucial to the smooth execution of this handover. Together, these efforts have unlocked a new tier of raw computing potential for the Wits research community.

Shining a Spotlight on Collaboration

To mark this milestone, our team arranged media interviews to document the handover. These interviews, currently in post-production, will soon provide a behind-the-scenes look at how shared infrastructure and institutional collaboration are driving the future of South African research.

Why This Matters

This transfer provides Wits with the foundational "bricks" of a high-performance data centre. With these 34 units, researchers can:

• Scale up complex scientific simulations across multiple server nodes.
• Expand data processing capabilities for multidisciplinary projects.
• Increase reliability and networking speed through dedicated PowerConnect switches.

This collaborative transfer reflects CHPC’s ongoing commitment to providing South Africa’s brightest minds with the robust infrastructure they need to innovate, while supporting student-led community engagement and workforce development.

What’s Next? Stay tuned for the full media coverage of this exciting chapter for Wits and CHPC! Together, we’re powering progress, one server at a time.

Continue Reading

Big News for CPUT: A New Era of Discovery Begins!

Big News for CPUT: A New Era of Discovery Begins!

March 28, 2026   Post,

Cape Peninsula University of Technology (CPUT) has just unlocked a new level of research and innovation with the arrival of the Stampede 1 OpenHPC Cluster. As the newest member of Africa’s expansive HPC Ecosystems project community, CPUT is poised to accelerate its contributions to scientific advancement and multidisciplinary collaboration.

This is more than just a technological upgrade; it’s a game-changer for CPUT’s researchers and students. Access to high-performance computing opens the door to ambitious projects, powerful partnerships, and hands-on experiences that can drive real impact across countless fields.

The successful deployment of Stampede 1 is thanks to the dedication and expertise of both CPUT’s technical team and CHPC’s HPC specialists, who worked together to ensure seamless integration into the broader ecosystem.

We are delighted to welcome CPUT into this dynamic and ever-growing network of innovators. Each new addition strengthens the HPC Ecosystems project community, bringing new perspectives, capabilities, and momentum to the South African and continental research landscape.

Congratulations to everyone who made this milestone possible. The future at CPUT is bright, and we look forward to the discoveries and breakthroughs that will follow!

Continue Reading

The successful delivery and commissioning of the Stampede 1 OpenHPC Cluster at the University of Mpumalanga (UMP)

The successful delivery and commissioning of the Stampede 1 OpenHPC Cluster at the University of Mpumalanga (UMP)

March 23, 2026   Post,

We are excited to share a key milestone in the CHPC HPC Ecosystems project: the successful delivery and commissioning of the Stampede 1 OpenHPC Cluster at the University of Mpumalanga (UMP). This accomplishment reflects exceptional teamwork across provinces and institutions and marks an important advance in strengthening South Africa’s national research infrastructure.

Seamless Deployment and Setup

Deployment began with the assembly, configuration, and thorough testing of the cluster at CHPC. Once the system met all quality checks, it was carefully transported to UMP. There, Sam Mathekga oversaw the unboxing, on-site setup, and network integration, ensuring a smooth transition to the university's environment and full operational readiness.

First Tests and Pilot Users

After installation, Sam Mathekga conducted site acceptance tests to confirm the cluster’s performance and stability within UMP’s operational environment. The successful outcome marked a major step toward project handover. Over the next six months, we, the CHPC HPC Ecosystems team, will facilitate ongoing support and hands-on training for UMP’s cluster administrators. During this period, a dedicated group of pilot users will rigorously test the system, pushing it to its limits to ensure peak reliability and readiness before the official launch. This collaborative, phased approach ensures the cluster will be fully embedded in the university’s research ecosystem from day one.

Empowering Research at the University of Mpumalanga

The deployment of the Stampede 1 OpenHPC Cluster represents a leap forward for the University of Mpumalanga’s research capabilities. With this new resource, researchers and students will have access to powerful HPC infrastructure, enabling advanced computational projects across various disciplines.

This milestone was achieved through the dedication and expertise of many. Special recognition goes to Sam Mathekga for leading the on-site deployment, Mfundo Mdwadube and Funeka Mafani (CHPC), Prof Thenjiwe Meyiwa (Vice-Chancellor & Principal), Prof Aldo Stroebel (DVC: Research, Innovation and Internationalisation), Joel Vele (CIO), Prof Ntombovuyo.Wayi-Mgwebi (Associate Professor in Information Communication Technology), Michael Ramatapa (Deputy Director ICT Operations) and his team for their pivotal roles in facilitating delivery and cluster development team, Mfundo Mdwadube, and Bryan Johnston (CHPC), with Bryan Johnston serving as project lead. We also gratefully acknowledge Bryan Johnston and Werner Janse van Rensburg for their project leadership and management, championing the initiative from inception to completion. Our sincere thanks extend to all colleagues at CHPC and UMP whose knowledge and commitment made this deployment possible.

This is just the beginning. As the Stampede 1 OpenHPC Cluster is put through its paces and new users come on board, we look forward to seeing innovative research and collaboration thrive at UMP. Stay tuned for further updates as the CHPC HPC Ecosystems project continues to grow and empower South African research.

​

​

Continue Reading

Reviving Legacy Servers for Modern HPC: Secure KVM Access with Dockerized Firefox-IcedTea

Reviving Legacy Servers for Modern HPC: Secure KVM Access with Dockerized Firefox-IcedTea

March 19, 2026   Post,

An Opinion Piece by Mfundo Mdwadube. 

---

Special thanks to the Texas Advanced Computing Centre (TACC) for providing us with Dell PowerEdge C8220 and C8000 hardware, and to the CHPC-CSIR for their ongoing support. This solution builds upon the excellent Docker image maintained by Rastislav Vojčík; his work on the Firefox-IcedTea container eliminated the need for insecure local Java installations, making this project possible.

Every HPC builder and/or operator knows the scenario: yesterday’s powerhouse hardware, still performing reliably, suddenly becomes a management headache when critical tools break with the march of progress. Decommissioning servers isn’t just about hardware failure or diminished performance. More often, it’s the creeping loss of software support, especially for crucial features like remote KVM access. As vendors drop support for legacy systems, those of us who maintain and repurpose older nodes are forced to get creative to keep them online, secure, and productive.

My team and I recently faced this dilemma with a rack of Dell PowerEdge C8220 servers, veterans of the “stampede1” era. The iDRAC 6 KVM tool, once essential for remote console access, now presented a maze of compatibility issues: unsupported Java applets, deprecated plugins, and browsers that refused to cooperate. Each workaround exposed new vulnerabilities or simply failed to launch, jeopardising our ability to manage and repurpose these still-capable nodes.

The breakthrough arrived when a senior colleague introduced me to the Firefox-IcedTea container. Building on this, I created an installer script that deploys a Dockerized Firefox browser bundled with IcedTea-Web. This setup recreated the vintage environment needed for iDRAC 6, without risking workstation or cluster security. By isolating vulnerable legacy code, providing seamless desktop integration, and leveraging modern tools (like SSH tunnelling) to securely bridge networks, we were able to restore full KVM functionality in just a few streamlined steps, without sacrificing usability or exposing systems to risk.

For HPC practitioners, the lesson is clear: containerization isn’t just about deploying the latest applications; it’s also a lifeline for legacy hardware. By encapsulating obsolete environments, containers extend the productive life of valuable infrastructure, enabling organisations to maximise their return on investment and maintain agility as hardware ages.

Don’t consign valuable hardware to the scrap heap. With container technology, open collaboration, and a bit of ingenuity, even decade-old servers can earn a second life in today’s demanding compute environments.

Continue Reading

Open HPC Virtual Lab #13: Special Edition

Open HPC Virtual Lab #13: Special Edition

March 18, 2026   Training,

 We’re excited to announce Virtual Lab #13: Special Edition, a focused, hands-on High-Performance Computing (HPC) training with OpenHPC 2.x, designed especially for late starters and those who need more time to complete the last session.

🗓️ Event Dates: March 17 – April 3, 2026

💻 Format: Online, self-paced, with dedicated office hours and expert support
🔗 Register now: https://events.chpc.ac.za/event/170/

What’s different about this edition?

• Extra support and flexibility for those catching up or starting late

• Practical, guided experience with OpenHPC 2.x

• Build your own virtual HPC cluster and gain skills you can use immediately

• Connect with peers and HPC experts

Who should join?

• Anyone who registered late or couldn’t finish Virtual Lab #12

• Newcomers eager for a hands-on, supportive start in HPC

• Researchers, engineers, and IT professionals looking to build practical HPC skills

Limited places available, register early to secure your spot!

Take the next step in your HPC journey with us. Join the Open HPC Virtual Lab #13! #HPC #OpenHPC #VirtualLab #CHPC #HPCtraining #CareerDevelopment

Continue Reading

OpenHPC Virtual Lab 12th Edition: That’s a Wrap!

OpenHPC Virtual Lab 12th Edition: That’s a Wrap!

March 16, 2026   Training,

Today, we officially conclude our latest virtual class! It was a true honour to co-host and lead this training once again alongside my colleague Bryan Jonston.

We had an incredible turnout with an eager group of students joining us from India, across Africa, and around the world. It’s been inspiring to watch everyone master OpenHPC at their own pace; some successfully smashed the virtual lab on time, while others are still diving deep to refine their knowledge.

A huge thank you to everyone who participated and made this session so dynamic! We look forward to seeing what this new cohort of HPC specialists achieves.

Continue Reading