Chaosblade

A single command-line tool that injects faults across hosts, the JVM, C++, containers, and Kubernetes by dispatching versioned YAML experiment specs to per-domain executor binaries.

• Category: Chaos Engineering
• CNCF maturity: Sandbox
• Language: Go (the blade CLI; a separate Python AI layer lives under blade-ai/)
• License: Apache-2.0
• Repository: chaosblade-io/chaosblade
• Documented at commit: 39a0c02 (2026-06-18, near tag blade-ai-v0.5.0)

What it is

Chaosblade is a chaos engineering tool from Alibaba. Chaos engineering is the practice of injecting controlled faults into a system to learn how it behaves under failure. The core deliverable is blade, a Go command-line interface (CLI) that creates, queries, and destroys fault-injection experiments such as CPU load, network latency, disk fill, and process kill.

The design centre is an experiment model: every scenario is a target plus an action plus flags, defined in versioned YAML rather than hard-coded. blade itself does not inject faults. It parses the YAML, builds an experiment record, and shells out to a separate executor binary (for example chaos_os) that does the real work. That separation lets one CLI cover hosts, the Java Virtual Machine (JVM), C++, Docker, the Container Runtime Interface (CRI), Kubernetes, and cloud providers without the CLI knowing each implementation.

State is kept in a local SQLite file, so a single host needs no external database to track running experiments and recover them.

When to use it

• You want one consistent CLI and experiment grammar to inject faults across heterogeneous targets (bare host, JVM application, container, Kubernetes pod) rather than a different tool per layer.
• You need application-layer fault injection, such as JVM method exceptions or C++ line-level faults, that a network or node-level tool cannot reach.
• You want experiments to self-recover: a timeout flag schedules an automatic destroy so a forgotten experiment does not linger.
• It is a weaker fit if you want a fully declarative, Kubernetes-native, custom-resource-driven workflow as the primary interface. For that the operator (chaosblade-operator) or alternatives like Chaos Mesh and LitmusChaos sit closer to the cluster.

In this deep-dive

• History: origin, milestones, and why it exists.
• Architecture: components and how requests flow.
• Adoption & Ecosystem: who runs it and what surrounds it.
• Internals: the code paths that matter, read from source.
• Getting Started: install and a first working setup.

Sources

1. chaosblade-io/chaosblade repository and README: https://github.com/chaosblade-io/chaosblade
2. Chaosblade CNCF project page (Sandbox, accepted 2021-04-28): https://www.cncf.io/projects/chaosblade/
3. ChaosBlade, An Open-Source Chaos Engineering Tool by Alibaba: https://www.alibabacloud.com/blog/chaosblade---an-open-source-chaos-engineering-tool-by-alibaba_594850
4. ChaosBlade: From the Chaos Engineering Experiment Tool to the Chaos Engineering Platform: https://www.alibabacloud.com/blog/chaosblade-from-the-chaos-engineering-experiment-tool-to-the-chaos-engineering-platform_598663
5. ChaosBlade-Box, a New Version of the Chaos Engineering Platform Has Released: https://chaosblade.io/en/blog/2022/06/24/ChaosBlade-Box-a-New-Version-of-the-Chaos-Engineering-Platform-Has-Released/
6. ChaosBlade documentation: https://chaosblade.io/en/docs/
7. LFX Insights, ChaosBlade: https://insights.linuxfoundation.org/project/chaosblade
8. Chaos Engineering in the Wild: Findings from GitHub (arXiv 2505.13654): https://arxiv.org/html/2505.13654
9. Local clone at commit 39a0c02e5f34af980f561440c0f1c218a3cde821: https://github.com/chaosblade-io/chaosblade/tree/39a0c02e5f34af980f561440c0f1c218a3cde821