Note:
This readme and related documentation are Work in Progress.
Chaos Mesh® is a cloud-native Chaos Engineering platform that orchestrates chaos on Kubernetes environments. At the current stage, it has the following components:
- Chaos Operator: the core component for chaos orchestration. Fully open sourced.
- Chaos Dashboard: a Web UI for managing, designing, monitoring Chaos Experiments; under development.
See the following demo video for a quick view of Chaos Mesh:
Chaos Operator injects chaos into the applications and Kubernetes infrastructure in a manageable way, which provides easy, custom definitions for chaos experiments and automatic orchestration. There are three components at play:
Controller-manager: used to schedule and manage the lifecycle of CRD objects
Chaos-daemon: runs as daemonset with privileged system permissions over network, Cgroup, etc. for a specific node
Sidecar: a special type of container that is dynamically injected into the target Pod by the webhook-server, which can be used for hijacking I/O of the application container
Chaos Operator uses Custom Resource Definition (CRD) to define chaos objects. The current implementation supports six types of CRD objects for fault injection, namely PodChaos, NetworkChaos, IOChaos, TimeChaos, StressChaos, and KernelChaos, which correspond to the following major actions (experiments):
- pod-kill: The selected pod is killed (ReplicaSet or something similar may be needed to ensure the pod will be restarted).
- pod-failure: The selected pod will be unavailable in a specified period of time.
- container-kill: The selected container is killed in the selected pod.
- netem chaos: Network chaos such as delay, duplication, etc.
- network-partition: Simulate network partition.
- IO chaos: Simulate file system faults such as I/O delay, read/write errors, etc.
- time chaos: The selected pod will be injected with clock skew.
- cpu-burn: Simulate the CPU of the selected pod stress.
- memory-burn: Simulate the memory of the selected pod stress.
- kernel chaos: The selected pod will be injected with (slab, bio, etc) errors.
See Docs.
This is a list of known adopters of Chaos Mesh. Some have already gone into production, and others are at various stages of testing.
See FAQs.
- Chaos Mesh - Your Chaos Engineering Solution for System Resiliency on Kubernetes
- Run Your First Chaos Experiment in 10 Minutes
- Simulating Clock Skew in K8s Without Affecting Other Containers on the Node
See the contributing guide and development guide.
Please reach out for bugs, feature requests, and other issues via:
- Following us on Twitter at @chaos_mesh.
- Joining the #project-chaos-mesh channel in the CNCF Slack workspace.
- Filing an issue or opening a PR against this repository.
On the fourth Thursday of every month (unless otherwise specified), the Chaos Mesh community holds a monthly meeting by video conference to discuss the status of Chaos Mesh.
Quick links:
- Integrate your Reliability Toolkit with Your World, Part 2
- Streake: Chaos Mesh によるカオスエンジニアリング
- Chaos Mesh : un générateur de chaos pour Kubernetes
See ROADMAP.
Chaos Mesh is licensed under the Apache License, Version 2.0. See LICENSE for the full license text.
Chaos Mesh® is a trademark of Beijing PingCap Xingchen Technology and Development Co., Ltd. © 2020 Beijing PingCap Xingchen Technology and Development Co., Ltd. All rights reserved.