selfhost.directory

The index of self-hostable projects — ranked, categorized and served with live updates.
Own your data and stop giving it to big tech.

GitHubWebsite
Chaos Mesh logo

Self-host Chaos Mesh

v2.8.7

Kubernetes chaos engineering platform for fault injection

ReleaseOpen SourceApache-2.0
GitHubWebsite
HelmFull install guide
helm install chaos-mesh chaos-mesh/chaos-mesh -n=chaos-mesh --create-namespace
Updated10 hours ago
LanguageGo
Setup difficultyAdvanced
Stars
7 956
Ports
2333 (Chaos Dashboard HTTP API and web UI), 2334 (Chaos Dashboard metrics endpoint), 31767 (Chaos Daemon gRPC server), 31766 (Chaos Daemon HTTP server), 10250 (Admission webhooks served by chaos-controller-manager), 10080 (Prometheus metrics exposed by chaos-controller-manager), 53 (chaos-dns-server DNS listener (DNSChaos))
Requires
Docker, containerd, or CRI-O container runtime on the nodes
Chaos Dashboard token login window with the link to open the token generatorChaos Dashboard token generator for creating user accounts and binding permissionsChaos Dashboard login form where a token and name are submitted

What is Chaos Mesh?

Chaos Mesh is an open-source, cloud-native chaos engineering platform for Kubernetes, used by teams that want to rehearse failures before they happen. Experiments are declared as Kubernetes custom resources covering pod, network, DNS, HTTP, I/O, time, stress, kernel, block-device, JVM, physical-machine and cloud (AWS, Azure, GCP) faults, and can be chained with Schedule, Workflow and StatusCheck resources for recurring or serial/parallel scenarios. Three runtime components do the work: the Chaos Controller Manager watches the CRDs and validates requests through admission webhooks, the Chaos Daemon runs as a privileged DaemonSet on nodes for container-, network- and kernel-level operations, and the optional Chaos Dashboard serves the HTTP API and web UI on port 2333. RBAC-based permission authentication with tokens is on by default, and GCP OAuth or generic OIDC login can be enabled through the Helm chart; multi-cluster injection works via the RemoteCluster resource, and the experimental Chaosd tool extends fault injection to physical machines. You install it with helm repo add chaos-mesh https://charts.chaos-mesh.org plus helm install, and it is a CNCF incubating project licensed under Apache-2.0. Dashboard data defaults to SQLite (MySQL and PostgreSQL are supported alternatives), so persistence must be configured explicitly or experiment history is lost on restart.

Should you self-host Chaos Mesh?

Chaos Mesh scores 69/100 on selfhost.directory (how scoring works) and ranks #2 of 8 in DevOps. It has 7 956 GitHub stars.

It shipped 5 stable releases in the past year. The latest, v2.8.7, came out on 28 October 2026. New versions land roughly every 6 days.

Setup is advanced: plan for several services and some ongoing maintenance. It is Apache-2.0-licensed and written mostly in Go.

Its closest alternative on selfhost.directory is Gitea Runner (74/100, 58 740 stars). It scores higher overall, so compare the two before you pick. See Chaos Mesh vs Gitea Runner side by side.

Features

  • Admission webhook validation
  • Application health checks (StatusCheck)
  • Broad fault injection coverage
  • Cloud chaos simulation
  • High-availability controller
  • Kubernetes-native chaos API (CRDs)
  • Multi-cluster fault injectionRemoteCluster is in an early stage.
  • Node-level chaos daemon (DaemonSet)
  • Physical-machine chaos toolkit (Chaosd)Marked experimental by the project.
  • Prometheus metrics export
  • RBAC-based user permissions
  • Recurring experiments (Schedule)
  • Target selection rules and namespace scoping
  • Web dashboard with HTTP API
  • Workflow Orchestration

Chaos Mesh README

How to install Chaos Mesh

Install Chaos Mesh with Helm

helm repo add chaos-mesh https://charts.chaos-mesh.org
helm install chaos-mesh chaos-mesh/chaos-mesh -n=chaos-mesh --create-namespace

Runtime-specific values must be set for containerd, K3s, MicroK8s or CRI-O clusters; an offline installation guide is also provided.

How to configure Chaos Mesh

  • CLEAN_SYNC_PERIOD12hSync period to clean up archived data

    Sync period to clean up archived data

  • clusterScopedtrueWhether Chaos Mesh manages cluster-wide chaos (see rbac.create and controllerManager.serviceAccount)

    Whether Chaos Mesh manages cluster-wide chaos (see rbac.create and controllerManager.serviceAccount)

  • DATABASE_DATASOURCEDatabase DSN used for Chaos Dashboard (deprecated in favour of dashboard.databaseSecretName)

    Database DSN used for Chaos Dashboard (deprecated in favour of dashboard.databaseSecretName)

  • DATABASE_DRIVERsqlite3Database driver for Chaos Dashboard data (SQLite is the default)

    Database driver for Chaos Dashboard data (SQLite is the default)

    values: sqlite3, MySQL, PostgreSQL

  • images.registryghcr.ioGlobal container registry for the images (can be replaced with a self-hosted registry)

    Global container registry for the images (can be replaced with a self-hosted registry)

  • LISTEN_HOST0.0.0.0Address chaos-dashboard listens on

    Address chaos-dashboard listens on

  • LISTEN_PORT2333Port chaos-dashboard listens on

    Port chaos-dashboard listens on

  • METRIC_PORT2334Port of the chaos-dashboard metrics endpoint

    Port of the chaos-dashboard metrics endpoint

  • METRICS_PORT10080Port on which chaos-controller-manager exposes Prometheus metrics

    Port on which chaos-controller-manager exposes Prometheus metrics

  • rbac.createtrueCreate the RBAC API objects

    Create the RBAC API objects

  • timezoneUTCTimezone used by controller-manager, chaos-daemon and dashboard (set on the TZ environment variable)

    Timezone used by controller-manager, chaos-daemon and dashboard (set on the TZ environment variable)

  • TTL_EVENT168hTTL of archived event data in Chaos Dashboard

    TTL of archived event data in Chaos Dashboard

  • TTL_EXPERIMENT336hTTL of archived experiment data

    TTL of archived experiment data

  • TTL_SCHEDULE336hTTL of archived schedule data

    TTL of archived schedule data

  • TTL_WORKFLOW336hTTL of archived workflow data

    TTL of archived workflow data

  • WEBHOOK_PORT10250Port on which chaos-controller-manager provides admission webhooks

    Port on which chaos-controller-manager provides admission webhooks

Chaos Mesh releases & version history

Even more releases RSS

FAQ

How can I persist Chaos Dashboard data?

The dashboard stores data in SQLite by default, which is lost on restart without a Persistent Volume; enable PV or switch the database engine to MySQL or PostgreSQL.

How can I disable the safe mode?

Safe mode (dashboard authentication) is on by default; set dashboard.securityMode to false at install or upgrade time, which should only be done outside production.

Failed to install Chaos Mesh with the message: no matches for kind "CustomResourceDefinition" in version "apiextensions.k8s.io/v1"?

This happens on Kubernetes v1.15 or older; create the v1beta1 CRDs manually with --validate=false and pass --skip-crds to Helm, or upgrade the cluster.

While trying to run DNSChaos in OpenShift, authorization problems blocked the process?

Grant the privileged Security Context Constraint to the chaos-dns-server service account.

While trying to install Chaos Mesh in OpenShift, authorization problems blocked the installation process?

Add the privileged Security Context Constraint to the chaos-daemon service account in the chaos-mesh namespace.

The default Google Cloud administrator account cannot create chaos experiments. How do I fix it?

That account cannot be checked by AdmissionReview, so create a ClusterRole with the needed chaos-mesh.org permissions and bind it to your Google Cloud user email.

If I have not deployed a Kubernetes cluster, can I use Chaos Mesh to create chaos experiments?

No — Chaos Mesh needs a Kubernetes cluster; without one you can use chaosd to inject failures instead.

More in DevOps

Rate this project

Click to rate — no account needed.

0 comments

Posting as guest
Loading comments…