chaostoolkit.org

What It Is

Chaos Toolkit is a chaos engineering toolkit for developers, designed to “explore and test systems to discover weaknesses.” Its core idea is Chaos as Code: experiments are declared in JSON/YAML, while steady-state hypotheses, probes, actions, and rollback steps are stored, collaborated on, and orchestrated like code.

Core Capabilities

The tool runs around the Experiment specification. An experiment includes a title, description, and method, and can also define a steady-state-hypothesis and rollbacks. The steady-state hypothesis queries system status through probes and uses tolerance to determine whether expectations are met. If the steady state is not satisfied before the experiment, the method steps are aborted. tolerance supports formats such as boolean, integer, string, list, range, regex, jsonpath, and probe. The Journal specification records the experiment, status, start and end times, duration, whether a deviation occurred, executed steps, and rollback results, making postmortems easier.

Language, Extensions, and Ecosystem

Experiment definitions support JSON/YAML, with providers including Python, HTTP, and Process. Its ecosystem is broad, covering AWS, Azure, GCP, Kubernetes, Ansible, Service Fabric, Spring, Kafka, ToxiProxy, Istio, WireMock, k6, Slack, DataDog, Dynatrace, Grafana, Honeycomb, OpenTelemetry, Prometheus, and more. It also emphasizes Open API and extensibility, allowing developers to build Driver Extensions and Control Extensions, and to embed it into CI/CD workflows.

Pricing and Deployment

The main documentation clearly states that Chaos Toolkit is an Apache 2-licensed open-source project, with no lock-in and community-driven development. Deployment options include local binaries, Docker, AWS ECS Task, Google Cloud Action, GitHub Action, GitLab Component, Kubernetes Operator, and more. The page mentions sponsor Reliably, which can run experiments from the cloud or elsewhere, but it does not disclose commercial pricing.

Pros, Cons, and Who It’s For

Its strengths are openness, code-based workflows, clear specifications, and a rich extension ecosystem. It is especially suitable for SRE, DevOps, platform engineering, and backend teams that want to integrate resilience validation into their pipelines. The limitation is that chaos engineering itself has a meaningful learning curve: teams need to design proper steady states, probes, and rollbacks. The main content does not show details on enterprise support, permissions, auditing, or a managed control console.

Access from China

The main documentation does not provide information about access from mainland China, payment, or mirrors, so access status is marked as unknown. If alternatives are needed, consider evaluating Gremlin, LitmusChaos, Chaos Mesh, or cloud provider fault-injection services.