Flipkart's Chaos Engineering Revolution: Insights from KubeCon + CloudNativeCon India 2026
In today's fast-paced digital landscape, system resilience is paramount. Chaos engineering addresses this by intentionally introducing faults to uncover weaknesses in your infrastructure. Flipkart's Central Reliability Engineering team has taken this discipline to the next level by building a centralized chaos platform on top of LitmusChaos, an open-source chaos engineering platform designed to help teams identify vulnerabilities in their systems.
The implementation at Flipkart includes several key customizations. They developed a hybrid multi-tenancy architecture to support various teams and workloads. A DaemonSet-based high-availability model for chaos injection ensures that faults can be introduced consistently across the cluster. Additionally, they created a Script Runner fault that allows for dynamic target selection and context chaining, enhancing the flexibility of their chaos experiments. For non-Kubernetes workloads, they also integrated a hybrid VM chaos extension, ensuring that their chaos engineering practices extend beyond just containerized environments.
In production, it's crucial to understand the implications of chaos engineering. While the benefits are significant, introducing faults can lead to unexpected behavior if not managed carefully. Ensure that your team is prepared for the potential fallout and has robust monitoring in place to observe the effects of injected chaos. The official docs don't call out specific anti-patterns here. Use your judgment based on your scale and requirements.
Key takeaways
- →Implement a hybrid multi-tenancy architecture for chaos engineering to support diverse workloads.
- →Utilize a DaemonSet-based model for high-availability chaos injection across your Kubernetes cluster.
- →Leverage Script Runner faults for dynamic target selection and context chaining in your chaos experiments.
- →Extend chaos engineering practices to non-Kubernetes workloads with a hybrid VM chaos extension.
Why it matters
By adopting chaos engineering, teams can proactively identify and address weaknesses, leading to more resilient systems and improved user experiences. This approach minimizes downtime and enhances overall service reliability.
When NOT to use this
The official docs don't call out specific anti-patterns here. Use your judgment based on your scale and requirements.
Want the complete reference?
Read official docsUnified observability — logs, uptime monitoring, and on-call in one place. Used by 50,000+ engineering teams to ship faster and sleep better.
Try Better Stack free →Building a Custom Metrics Exporter for Kubernetes: A Practical Guide
Custom metrics exporters are essential for monitoring application states in Kubernetes. By exposing metrics through a simple HTTP server, you can gain insights into your application's performance. Learn how to implement this with concrete examples and avoid common pitfalls.
Diagnosing Kubernetes Control Plane Performance with AWS DevOps Agent
Kubernetes control plane performance can make or break your cluster's stability. The AWS DevOps Agent autonomously identifies issues, correlating CloudWatch logs with throttling patterns to deliver actionable insights. This article dives into how to leverage this tool effectively in production environments.
Speed Up Your Volcano Workload Insights with Headlamp
Tired of slow inspections of your Volcano workloads? Headlamp integrates seamlessly with Volcano, allowing you to visualize workload states and queue behaviors in one place. Dive into the specifics of how this integration enhances your Kubernetes experience.
Get the daily digest
One email. 5 articles. Every morning.
No spam. Unsubscribe anytime.