Break
Your System
Constructively.

Chaos Mesh brings various types of fault simulation to Kubernetes and has an enormous capability to orchestrate fault scenarios. It helps you conveniently simulate various abnormalities that might occur in reality during the development, testing, and production environments and find potential problems in the system.

Get Started →GitHub

Try it out with the following command 👇

Users of Chaos Mesh

Make Cloud Native + Chaos Engineering simple and straightforward.

Based on the principles of Chaos Engineering, Chaos Mesh abstracts real-world events into objects that can be directly applied, hiding the trivial details.

Design for Kubernetes

In the Kubernetes realm, CustomResourceDefinition (CRD) is a proven solution for implementing custom resources. Chaos Mesh leverages CRDs for natural integration with the Kubernetes ecosystem.

Easy to Use

No special dependencies, Chaos Mesh can be easily deployed on Kubernetes clusters directly, including minikube and kind.

Ability to perform chaos experiments in production environments without modifying the deployment logic of applications.
Quickly create chaos experiments through the dashboard, allowing users to observe the experiment's state in real time and quickly roll back any injected failures.

Flexible Scope

Chaos Mesh provides multiple filtering rules that allow selecting injection targets based on labels, annotations, and so on. They are collectively called selectors.

In addition, users can also set a namespace whitelist to greater control the "blast radius" of the experiment.

Security First

Role-based access control (RBAC) is enabled by default to restrict malicious operations. Users must be authorized with specific privileges to access the dashboard.

Wide variety of failure types.

Chaos Mesh initially started from a distributed system perspective, fully considering its possible failures, thus providing more comprehensive and fine-grained fault types to help users with fault injection for networks, disks, file systems, operating systems, etc.

Explore More

PodChaos / pod-failure

apiVersion: chaos-mesh.org/v1alpha1
kind: PodChaos
metadata:
  name: pod-failure-example
  namespace: chaos-mesh
spec:
  selector:
    labelSelectors:
      'environment': 'staging'
  mode: one
  action: pod-failure
  duration: 30s

NetworkChaos / delay

apiVersion: chaos-mesh.org/v1alpha1
kind: NetworkChaos
metadata:
  name: network-delay-example
spec:
  selector:
    namespaces:
      - default
    labelSelectors:
      'env': 'production'
  mode: one
  action: delay
  delay:
    latency: '10ms'
    correlation: '100'
    jitter: '0ms'

StressChaos

apiVersion: chaos-mesh.org/v1alpha1
kind: StressChaos
metadata:
  name: mem-stress
  namespace: chaos-mesh
spec:
  selector:
    namespaces:
      - 'default'
    labelSelectors:
      'env': 'test'
  mode: one
  stressors:
    memory:
      workers: 2
      size: '128MB'

Orchestrate complex fault scenarios with Workflows.

Real-world failures are often not isolated causes.Chaos Mesh has built-in workflows that allow you to experiment serially or in parallel at will to build walkthroughs that fit the architecture.

Suspend

You can also use the suspend node to simulate a temporary recovery.

Status Check

You can also use customized status checks to inform the cluster status.

Start Creating →

Experimental

Meet Chaosd: A Chaos Toolkit for Physical Machines.

Even if you are not using Kubernetes, you can still take advantage of the features offered by Chaos Mesh. One of our experimental tools, Chaosd, is specifically designed to test chaos on physical machines. Moreover, you can use PhysicalMachineChaos in Chaos Mesh to remotely invoke Chaosd for conducting experiments on physical machines.

Overview →