Create a chaos experiment that uses an agent-based fault to add CPU pressure to a Linux VM with the Azure portal

You can use a chaos experiment to verify that your application is resilient to failures by causing those failures in a controlled environment. In this guide, you will cause a high CPU event on a Linux virtual machine using a chaos experiment and Azure Chaos Studio. Running this experiment can help you defend against an application becoming resource-starved.

These same steps can be used to set up and run an experiment for any agent-based fault. An agent-based fault requires setup and installation of the chaos agent, unlike a service-direct fault, which runs directly against an Azure resource without any need for instrumentation.


Enable Chaos Studio on your virtual machine

Chaos Studio cannot inject faults against a virtual machine unless that virtual machine has been onboarded to Chaos Studio first. You onboard a virtual machine to Chaos Studio by creating a target and capabilities on the resource, then installing the chaos agent. Virtual machines have two target types - one that enables service-direct faults (where no agent is required), and one that enabled agent-based faults (which requires the installation of an agent). The chaos agent is an application installed on your virtual machine as a virtual machine extension that allows you to inject faults in the guest operating system.

Install stress-ng

The Chaos Studio agent for Linux requires stress-ng, an open-source application that can cause various stress events on a virtual machine. You can install stress-ng by connecting to your Linux virtual machine and running the appropriate installation command for your package manager, for example:

sudo apt-get update && sudo apt-get -y install unzip && sudo apt-get -y install stress-ng


sudo dnf -y install && sudo yum -y install stress-ng

Enable chaos target, capabilities, and agent


Prior to completing the steps below, you must create a user-assigned managed identity and assign it to the target virtual machine or virtual machine scale set.

  1. Open the Azure portal.
  2. Search for Chaos Studio (preview) in the search bar.
  3. Click on Targets and navigate to your virtual machine. Targets view in the Azure portal
  4. Check the box next to your virtual machine and click Enable targets then Enable agent-based targets from the dropdown menu. Enabling targets in the Azure portal
  5. Select the Managed Identity that you will use to authenticate the chaos agent and optionally enable Application Insights to see experiment events and agent logs. Selecting a managed identity
  6. Click Review + Enable then click Enable. Reviewing agent-based target enablement
  7. After a few minutes, a notification will appear indicating that the resource(s) selected were successfully enabled. The Azure portal will add the user-assigned identity to the virtual machine, enable the agent target and capabilities, and install the chaos agent as a virtual machine extension. Notification showing target successfully enabled
  8. If enabling a virtual machine scale set, upgrade instances to the latest model by going to the virtual machine scale set resource blade, clicking Instances, then selecting all instances and clicking Upgrade if not on the latest model.

You have now successfully onboarded your Linux virtual machine to Chaos Studio. In the Targets view you can also manage the capabilities enabled on this resource. Clicking the Manage actions link next to a resource will display the capabilities enabled for that resource.

Create an experiment

With your virtual machine now onboarded, you can create your experiment. A chaos experiment defines the actions you want to take against target resources, organized into steps, which run sequentially, and branches, which run in parallel.

  1. Click on the Experiments tab in the Chaos Studio navigation. In this view, you can see and manage all of your chaos experiments. Click on Add an experiment Experiments view in the Azure portal
  2. Fill in the Subscription, Resource Group, and Location where you want to deploy the chaos experiment. Give your experiment a Name. Click Next : Experiment designer > Adding basic experiment details
  3. You are now in the Chaos Studio experiment designer. The experiment designer allows you to build your experiment by adding steps, branches, and faults. Give a friendly name to your Step and Branch, then click Add fault. Experiment designer
  4. Select CPU Pressure from the dropdown, then fill in the Duration with the number of minutes to apply pressure and pressureLevel with the amount of CPU pressure to apply. Leave virtualMachineScaleSetInstances blank. Click Next: Target resources > Fault properties
  5. Select your virtual machine, and click Next Add a target
  6. Verify that your experiment looks correct, then click Review + create, then Create. Review and create experiment

Give experiment permission to your virtual machine

When you create a chaos experiment, Chaos Studio creates a system-assigned managed identity that executes faults against your target resources. This identity must be given appropriate permissions to the target resource for the experiment to run successfully.

  1. Navigate to your virtual machine and click on Access control (IAM). Virtual machine overview page
  2. Click Add then click Add role assignment. Access control overview
  3. Search for Reader and select the role. Click Next Assigning Virtual Machine Contributor role
  4. Click Select members and search for your experiment name. Select your experiment and click Select. If there are multiple experiments in the same tenant with the same name, your experiment name will be truncated with random characters added. Adding experiment to role
  5. Click Review + assign then Review + assign.

Run your experiment

You are now ready to run your experiment. To see the impact, we recommend opening an Azure Monitor metrics chart with your virtual machine's CPU pressure in a separate browser tab.

  1. In the Experiments view, click on your experiment, and click Start, then click OK. Starting experiment
  2. When the Status changes to Running, click Details for the latest run under History to see details for the running experiment.

Next steps

Now that you have run an agent-based experiment, you are ready to: