Introduction to ChaosMeta: An Open‑Source Cloud‑Native Chaos Engineering Platform

ChaosMeta is a cloud‑native chaos engineering platform designed for automated fault‑injection experiments, originating from Ant Group's internal XMonkey platform and now open‑sourced to share the company’s extensive methodology, technical capabilities, and product features.

The platform provides a one‑stop solution covering the entire chaos engineering lifecycle, enabling rapid discovery of potential risks in business applications and systems, and includes a built‑in "risk catalog" that aggregates common technical risks across various domains.

ChaosMeta’s lifecycle model addresses common pain points such as admission checks, traffic injection, fault injection, fault measurement, recovery measurement, and post‑exercise analysis, offering technical support for each stage to facilitate automated chaos engineering.

Its rich cloud‑native fault‑injection capabilities allow scenarios like mass pending pods to overload the scheduler, dynamic webhook validation delays, field mutation via webhook, and heavy Watch&List connections that strain the API server, extending beyond typical system‑resource and network faults.

The platform architecture follows an Operator‑based design with three layers: a user layer (chaosmeta‑platform) providing a visual interface and low‑entry‑barrier usage; an engine layer delivering remote injection, orchestration, measurement, and cloud‑native fault capabilities; and a kernel layer (chaosmetad and chaosmeta‑daemonset) offering single‑node fault injection via HTTP service or CLI.

Future roadmap is divided into platform capabilities and fault‑injection capabilities, progressing through three stages: manual configuration, automation with risk‑catalog‑driven health‑check packages, and intelligent AI‑assisted risk scenario generation.

ChaosMeta encourages community participation through open‑source development, welcoming contributions, discussions, and feedback, with resources available on GitHub, official documentation, and community groups.