Chaos Mesh

Chaos Mesh is a free, open-source fault-injection platform for teams running Kubernetes. It is best suited to operators who want to manage experiments through cluster-native resources and cover faults across applications and infrastructure. Its breadth and orchestration are strong, but the privileged daemon and the care needed to contain experiments make it an operational choice, not a casual testing add-on.

Overview

Chaos Mesh uses Kubernetes CustomResourceDefinitions and the Kubernetes API to manage experiments in development, testing, and production. That approach suits teams already comfortable administering Kubernetes; it is a poor fit for organizations seeking a managed service or a tool independent of Kubernetes. Experiments can simulate faults affecting Pods, networks, DNS, HTTP, I/O, time, CPU and memory, kernels, block devices, JVMs, physical machines, and AWS, Azure, and GCP.

Selectors and namespace controls help target experiments and contain their blast radius. They do not remove the need to plan for recovery: for Pod Kill experiments, the project recommends a ReplicaSet or similar mechanism so the Pod can restart automatically. Chaos Mesh began as part of TiDB’s core testing platform, and is licensed under Apache License 2.0.

Key features

Scheduling, workflows, and health checks

Schedule resources support recurring experiments; Workflow resources can run experiments serially or in parallel and check their status and results. StatusCheck resources add application health checks. These controls make it possible to coordinate repeatable exercises rather than launch isolated faults, while giving teams a way to check application health during an experiment.

Dashboard and access

Chaos Dashboard provides a web interface and HTTP API for creating, managing, and observing experiments. Role-based access control is enabled by default, and Kubernetes RBAC and namespace annotations provide additional ways to restrict permissions and where experiments are enabled. On Google Cloud Platform, the dashboard can use Google OAuth authentication.

The dashboard uses SQLite by default. Without persistent volumes, its data is lost after a reboot; teams that need persistence can configure MySQL or PostgreSQL instead. That setup is a meaningful operational consideration for anyone relying on dashboard history or state.

Deployment and integrations

The official Helm repository supports Helm 3 and provides a command to install Chaos Mesh into a Kubernetes namespace. Chaos Mesh is self-hosted, with API and web access. Its Grafana data-source plugin requires Chaos Mesh 2.1 or later and Grafana 10.0 or later; it was tested on Grafana 10.0.3. The chaos-mesh-action GitHub Action can integrate experiments into GitHub Actions workflows.

RemoteCluster can install Chaos Mesh components in remote Kubernetes clusters and synchronize experiment status, but the feature is in an early stage. Teams should treat it as an emerging option rather than a mature basis for multi-cluster operations.

Pricing

Chaos Mesh is free and open source. The Chaos Mesh plan costs nothing and includes the Kubernetes-based platform; there is no paid tier or seat quota in the stated plan. The trade-off is self-hosting: teams operate the Kubernetes components, dashboard storage, and experiment safeguards themselves.

Compatibility depends on the release and Kubernetes version. Chaos Mesh 2.8 supports Kubernetes 1.30 through 1.35, while version 2.7 supports Kubernetes 1.26 through 1.28. Check that pairing before choosing a release. Community technical support is available through CNCF Slack’s #project-chaos-mesh channel and GitHub Discussions. Security issues are fixed as soon as possible and back-ported to the last two releases.

Platforms

Chaos Mesh runs on Kubernetes and is self-hosted, with API and web interfaces. Its cloud fault injection covers AWS, Azure, and GCP, but the platform itself remains Kubernetes-centered rather than a general-purpose desktop or hosted application.

Who it's for

Chaos Mesh is a strong fit for platform, SRE, and reliability teams that already operate Kubernetes and want to schedule, scope, and observe fault experiments through the Kubernetes API. Its broad fault coverage and workflow controls can support exercises spanning application and infrastructure behavior. Teams without Kubernetes expertise, those seeking vendor-managed operations, or those unwilling to grant a privileged component access to nodes should look elsewhere.

Pros and cons

Pros

• Broad fault coverage spans workloads, networking, time, kernels, JVMs, physical machines, and major cloud platforms.
• Kubernetes-native resources, recurring schedules, workflows, and health checks support repeatable, coordinated experiments.
• Selectors, namespace controls, and default role-based access control help teams restrict experiment scope and dashboard operations.
• The Apache-licensed platform is free to self-host, with Helm 3 installation and a GitHub Actions integration.

Cons

• Chaos Daemon runs as a privileged Kubernetes DaemonSet by default, a significant access decision for security-conscious operators, even though the privilege can be disabled.
• Dashboard data stored in default SQLite is lost on reboot without persistent volumes; MySQL or PostgreSQL requires configuration.
• RemoteCluster is described as early-stage, so it is less compelling for teams needing established remote-cluster management.
• Safe use requires deliberate scoping and recovery planning, especially for Pod Kill experiments.

Alternatives

Compare chaos engineering platforms if you want to assess the broader category. Chaos Toolkit is another free, Apache-licensed open-source option with API, desktop, self-hosted, and web platforms; consider it if those platform options better match your environment. ChaosBlade is free and describes more than 200 experiment scenarios across three system platforms and four programming-language applications, which may suit readers focused on that scenario breadth.

Reliably has a free plan for one team member and describes its open-source approach as avoiding lock-in; consider it if those terms are a better fit. Toxiproxy is a free, open-source TCP proxy under the MIT license, a narrower option for readers focused on TCP fault injection. Krkn is another free option for readers comparing chaos engineering tools. chaosd is a free Apache-2.0 option distributed as a Linux binary and requiring glibc v2.17 or later. FaultMesh offers a free Starter plan for up to three services with core fault types, basic resilience scoring, and community support; its paid Growth plan is $49 a month. Krkn-AI is a free, Apache License 2.0 option for Linux and self-hosted use.

Verdict

Choose Chaos Mesh if your team runs Kubernetes and wants free, Kubernetes-native experiments with broad fault coverage, scheduling, and controls for limiting scope. The main reason to choose it is that combination of breadth and cluster-native orchestration. Look elsewhere if you need managed operations, a non-Kubernetes-centered tool, or a less privileged deployment by default.