Chaos Runner is a lifecycle manager for chaos experiments that orchestrates the execution of LitmusChaos experiments on Kubernetes clusters.
The tool solves the problem of bridging communication and resource management between the Chaos Operator and individual chaos experiment jobs. It runs as a pod in the chaos namespace and handles the complete experiment lifecycle: reading chaos parameters from experiment custom resources, merging overrides from the ChaosEngine, validating dependencies like ConfigMaps and Secrets, constructing and launching experiment jobs, monitoring their execution, and cleaning up resources according to configured policies. The runner patches the ChaosEngine with experiment verdicts and creates events tracking different execution phases, enabling full audit logging from engine creation through removal.
Teams running LitmusChaos on Kubernetes should adopt this tool if they need reliable experiment orchestration with dependency validation and cleanup automation. It suits projects requiring traceable chaos experiment execution with support for batch runs, parallel execution, and proper garbage collection through owner references. The tool is particularly valuable when you need to abort experiments in progress and remove all chaos residue in a single operation, or when managing multiple experiment combinations with randomized execution patterns.
The project is in beta stage with further improvements planned. Development activity shows ongoing refinement of core functionality, with the team actively maintaining the codebase and welcoming community contributions through issue reporting, documentation improvements, and framework enhancements.
