netflix/chaosmonkey

Chaos Monkey is a resilience testing tool from Netflix that deliberately causes random failures in a running production system. It works by randomly terminating virtual machine instances and containers, the individual server processes that keep an application running, while the system is live. The goal is not to break things maliciously, but to force engineers to build services that can survive unexpected failures gracefully. If your system can handle Chaos Monkey randomly killing pieces of it, it is far less likely to collapse during an unplanned real-world outage. This approach is part of a broader discipline called Chaos Engineering, the practice of intentionally introducing controlled failures to expose weaknesses before they cause customer-facing problems. Chaos Monkey is written in Go and is designed to work with Spinnaker, a continuous delivery platform (a system for automatically deploying software updates). It integrates with various cloud backends including AWS, Google Compute Engine, Azure, Kubernetes, and Cloud Foundry. You need to be managing your applications through Spinnaker to use Chaos Monkey. You would use this tool if you are a reliability or infrastructure engineer at a company running large cloud-based services and you want to proactively test whether your system degrades gracefully when individual components fail.