Oldest Kubernetes version supported: 1.16
⚠️ Kubernetes version 1.20.x is not supported!This Kubernetes issue prevents the controller from running properly on Kubernetes 1.20.0-1.20.4. Earlier versions of Kubernetes as well as 1.20.5 and later are still supported.
💣 Disclaimer 💣
The Chaos Controller allows you to disrupt your Kubernetes infrastructure through various means including but not limited to: bringing down resources you have provisioned and preventing critical data from being transmitted between resources. The use of Chaos Controller on your production system is done at your own discretion and risk.
The Chaos Controller is a Kubernetes controller with which you can inject various systemic failures, at scale, and without caring about the implementation details of your Kubernetes infrastructure. It was created with a specific mindset answering Datadog's internal needs:
- 🐇 Be fast and operate at scale
- At Datadog, we are running experiments injecting and cleaning failures to/from thousands of targets within a few minutes.
- 🚑 Be safe and operate in highly disrupted environments
- The controller is built to be able to limit the blast radius of failures but also to be able to recover by itself in catastrophic scenarios.
- 💡 Be smart and operate in various technical environments
- With Kubernetes, all environments are built differently.
- Whatever your cluster configuration and implement details choice, the controller is able to inject failures by relying on low-level Linux kernel features such as cgroups, tc or even eBPF.
- 🪙 Be simple and operate at low cost
- Most of the time, your Chaos Engineering platform is waiting and doing nothing.
- We built this project so it uses resources only when it is really doing something:
- No DaemonSet or any always-running processes on your nodes for injection, no reserved resources when it's not needed.
- Injection pods are created only when it is needed, killed once experiment is done, and built to be evicted if necessary to free resources.
- A single long-running pod, the controller, and nothing else!
💡 Read the latest release quick installation guide and the configuration guide to know how to deploy the controller.
Disruptions are built as short-living resources which should be manually created and removed once your experiments are done. They should not be part of any application deployment. The Disruption resource is immutable. Once applied, you can't edit it. If you need to change the disruption definition, you need to delete the existing resource and to re-create it.
Getting started is as simple as creating a Kubernetes resource:
apiVersion: chaos.datadoghq.com/v1beta1kind: Disruptionmetadata:
name: node-failurenamespace: chaos-demo # it must be in the same namespace as targeted resourcesspec:
selector: # a label selector used to target resourcesservice: demo-curlcount: 1# the number of resources to target, can be a percentage if you suffix with "%", e.g. `count: 50%`duration: 1h# the amount of time before your disruption automatically terminates itself, for safetynodeFailure: # trigger a kernel panic on the target nodeshutdown: false # do not force the node to be kept downTo disrupt your cluster, run kubectl apply -f <disruption_file>.yaml. You can clean up the disruption with kubectl delete -f <disruption_file>.yaml. For your safety, we recommend you get started with the dry-run mode enabled.
📖 The features guide details all the features of the Chaos Controller.
📖 The examples guide contains a list of various disruption files that you can use.
Check out Chaosli if you want some help understanding/creating disruption configurations.
New feature in
8.0.0
The Chaos Controller has expanded its capabilities by introducing disruption scheduling, enhancing your ability to automate and test system resilience consistently. Instead of manual creation and deletion, use DisruptionCron to regularly disrupt long-lived Kubernetes resources like Deployments and StatefulSets.
apiVersion: chaos.datadoghq.com/v1beta1kind: DisruptionCronmetadata:
name: node-failurenamespace: chaos-demospec:
schedule: "*/15 * * * *"# every 15 minutestargetResource: kind: deploymentname: demo-curldisruptionTemplate:
count: 1duration: 1hnodeFailure:
shutdown: falseTo schedule disruption in your cluster, run kubectl apply -f <disruption_cron_file>.yaml. To stop, run kubectl delete -f <disruption_cron_file>.yaml.
🔎 Check out DisruptionCron guide for more detailed information on how to schedule disruptions.
Chaos Engineering is necessarily different from system to system. We encourage you to try out this tool, and extend it for your own use cases. If you want to run the source code locally to make and test implementation changes, visit the Contributing Doc. By the way, we welcome Pull Requests.