Lessons from My CNCF LFX Mentorship with Karmada

Krishiv Mahajan — Thu, 28 May 2026 05:17:33 +0000

Lessons from My CNCF LFX Mentorship with Karmada

A reflection on open-source contribution, real-world challenges, and the power of mentorship.

When I first applied for the Linux Foundation Mentorship (LFX) program, I had a rough idea of what I was getting into — open-source contribution, cloud-native technology, and a structured learning path. What I did not expect was just how much the experience would stretch me: technically, academically, and personally. This is my story.

The Project: Improving Karmada quick start

My mentorship was centered around Karmada, an open-source multi-cloud, multi-cluster Kubernetes management system. The project gave me the opportunity to contribute in meaningful ways to its developer experience and learning ecosystem.

Improving the Quick Start

One of my primary contributions was improving Karmada's Quick Start experience. The goal was straightforward in theory — make it easier for new users and contributors to get Karmada up and running with minimal friction. In practice, this involved understanding the entire bootstrapping flow, identifying pain points, and iterating on documentation ,improving scripts and adding automations to make the onboarding experience smoother and more reliable.

This kind of work may not sound glamorous, but it is one of the most impactful things you can do in an open-source project. A great quick start is often the difference between a developer adopting a technology or walking away from it.

The macOS CI Rabbit Hole

One of the more adventurous detours of my mentorship was diving deep into CI (Continuous Integration) pipeline support on macOS. The idea was to extend Karmada's CI coverage to macOS environments, which would broaden the project's reach and ensure it works well for Mac-based contributors.

I spent a significant amount of time researching this — reading through GitHub Actions documentation, studying macOS runner capabilities, and exploring how virtualization fits into automated testing workflows. It was a genuinely exciting research challenge.

However, I hit a hard wall: virtualization limitations in GitHub Actions. macOS runners on GitHub Actions have restrictions on nested virtualization, which is a core requirement for spinning up the kind of cluster environments Karmada needs for integration testing. Despite my best efforts to find workarounds, this constraint was ultimately a blocker the project couldn't circumvent at the time, and the task was formally dropped.

It was disappointing, but I want to be clear — it was far from wasted time. I came away with a much deeper understanding of:

How CI/CD pipelines are architected for cloud-native projects
The nuances of macOS virtualization and its current limitations in cloud environments
How to evaluate technical feasibility and communicate blockers clearly to mentors and the community

Sometimes the most valuable learning comes from the tasks that don't make it to the finish line.

The Killercoda Tutorials: Easier Said Than Done

Toward the end of the mentorship, I was tasked with building and refining interactive Killercoda tutorials for Karmada. When I first heard "tutorials," my initial reaction was something like: "Great, the easy part!"

I was wrong.

I worked on scenarios covering some of Karmada's most powerful features including:

Application-level Failover — demonstrating how Karmada automatically reschedules workloads when a cluster goes down
FederatedHPA (Horizontal Pod Autoscaling) — cross-cluster autoscaling based on aggregated metrics
Multi-Component Workload Scheduling — atomically scheduling complex, multi-resource workloads like Apache Flink deployments across clusters, using the MultiplePodTemplatesScheduling feature gate
HA Workload strategies — duplicated and divided scheduling for high-availability deployments

Each scenario required me to truly understand the feature — not just copy commands from documentation, but deeply grasp why each configuration decision was made, what could go wrong, and how to explain it clearly to a learner with no prior Karmada experience.

For example, building the Flink multi-component scheduling tutorial meant understanding leader election deadlocks in Karmada control plane components, leveraging Karmada's built-in resource interpreters for robust component population, and ensuring the tutorial was reproducible end-to-end in a constrained environment. That's not tutorial writing — that's systems engineering and technical writing at the same time.

The tutorials pushed me to understand the Karmada ecosystem at a depth I didn't anticipate going in, and I'm genuinely proud of what came out.

The Mid-Point Reality Check

I want to be honest: the middle of the mentorship was hard.

Everything started smoothly. I was enthusiastic, the tasks were well-defined, and momentum felt great. But as the weeks went on, the complexity of the work ramped up — and so did my university schedule. Exams, lab practicals, assignments, and project deadlines all converged at once.

Balancing an intensive open-source mentorship with a full academic load is not something I'd underestimate again. There were days I felt stretched thin, unsure if I could keep up the quality of contribution I wanted to deliver.

What made the difference? My mentors.

They were patient, understanding, and genuinely invested in my growth — not just as a contributor, but as a person. When I communicated my challenges, they helped me re-prioritize, set realistic expectations, and find ways to make progress even in smaller chunks. That kind of support is what separates a great mentorship program from just an internship.

The Finish Line

In the end, I made it through. I successfully passed the final LFX evaluations, and looking back, I can honestly say this mentorship has been one of the most formative experiences of my early career.

I contributed to a real-world, production-grade open-source project used by companies at scale. I navigated technical dead ends and learned to communicate them. I built educational content that will help future Karmada users get up to speed faster. And I proved to myself that I can handle the intersection of academic life and professional contribution — even when it's challenging.

Key Takeaways

Open source is collaborative, not competitive. The Karmada community was welcoming and supportive throughout.
Research has value even when it doesn't ship. My macOS CI investigation didn't land, but the knowledge did.
Documentation and tutorials are hard. They require deep technical understanding and clear communication — both skills worth developing.
Mentors matter more than tasks. The people who guide you shape your learning as much as the work itself.
Balance is a skill. Managing university and a mentorship simultaneously is challenging, but it's also the reality of early-career life. Learning to navigate it early is invaluable.

What's Next?

This mentorship has only deepened my interest in cloud-native technologies and the Kubernetes ecosystem. I plan to continue contributing to Karmada and stay engaged with the community. There's still so much to build.

If you're a student considering applying for LFX — do it. It will challenge you in ways you don't expect, and reward you in ways you can't fully anticipate until you're on the other side.

Thank you to my mentors, the Karmada community, and the Linux Foundation for this incredible opportunity.

DEV Community: Krishiv Mahajan

Lessons from My CNCF LFX Mentorship with Karmada

Lessons from My CNCF LFX Mentorship with Karmada

The Project: Improving Karmada quick start

Improving the Quick Start

The macOS CI Rabbit Hole

The Killercoda Tutorials: Easier Said Than Done

The Mid-Point Reality Check

The Finish Line

Key Takeaways

What's Next?