krkn-demos

This is a collection of scripts to easily demo krkn-hub scenarios. All the scenarios, cluster and cloud settings are centralized in the file config.yaml from where all the scripts gather the informations needed to run. The platform currently supported by the scripts is Openshift, if needed, without too much effort they could be ported to Kubernetes as well.

There are three categories of scripts:

• run scripts
  • the scripts needed to run the scenarios
• check scripts
  • the scripts needed to show live the effects of the respective scenarios
• service scripts
  • login.sh
    • performs the login in the OCP cluster
  • deploy_dittybopper.sh
    • deploys in the cluster an example workload that can be targeted from different scenario (which settings are the default in the config.yaml)

requirements

Warning

These script are currently compatible with Linux. MacOs users need to modify podman commands adding each environment variable as an argument of the podman command since the --env-host option is not supported on this platform. Few MacOS example scripts are available with the suffix macos_run*

To run these scripts, the following tools must be installed and available in the $PATH:

• yq yaml manipulation cli GitHub repo
• oc ocp CLI [GitHub repo]
• watch [man page]
• podman containerization platform by RedHat [website]
• curl command line http client [GitHub repo]
• aws Amazon Webservice CLI tool [GitHub repo]
• base64 base 64 encoder/decoder [man page]

Tip

For a better demo experience, it is highly recommended to install tmux [GitHub repo] and use the vertical or horizontal split to run both the run and check scripts simultaneously. To do this, start tmux, then press ctrl + b, followed by % to split the screen vertically, or " to split it horizontally. To navigate between panes, use ctrl + b and the appropriate arrow key.

cluster login

To login in the cluster the login.sh can be used. It supports both token and username/password login. To set the required parameters refer to the ocp section in the config.yaml. If the login succeeds a kubeconfig file will be written in the same folder.

Note

If a token value is provided, the username and password will be ignored. Ensure you set only one option and leave the other blank.

dittybopper deployment

Dittybopper is a comprehensive Prometheus/Grafana distribution that can be deployed on OpenShift, and is frequently used as the target for our pod and service disruption scenarios. To deploy it on the cluster, after successfully logging in, simply run the deploy_dittybopper.sh script. Once the workload has been deployed and is ready, the service URL will be displayed and can be configured in the config.yaml file under the various service-url settings for use by the check scripts.

All pod and service disruption scenarios are already configured to target this workload, so if it suits your demo, no further adjustments are necessary.

config.yaml

Each script gathers its parameters from this file, which should be considered the sole source for configuring scenario settings. The file is organized into the following sections:

ocp

Openshift credentials and API endpoint

aws

AWS credentials and region

scenarios

It includes global settings for the scenario, along with a sub-section containing scenario-specific configurations:

• app-outage
• node-outage
• pod-scenario
• pvc-scenario
• service-hijacking
• zone-outage

For an explanation of the specific scenario parameters, please refer to the available scenarios section below.

Some scenario-specific sub-sections include a check section, which is intended to hold parameters required for the corresponding check script to function. For example, the check section in the service-hijacking scenario contains the external URL of the target service, allowing real-time display of the hijacking results in the HTTP response.

available scenarios

The currently available scenarios, along with their corresponding check scripts, are:

• Application Outages (run_app_outage.sh, check_app_outage.sh)
• Node Failures (run_master_node_outage.sh, check_master_node_outage.sh)
• Pod Failures (run_pod_scenarios.sh, run_pod_scenarios.sh)
• PVC Disk Fill (run_pvc_scenario.sh, check_pvc_scenario.sh)
• Service Hijacking (run_service_hijacking.sh, check_service_hijacking.sh)
• Zone Outages (run_zone_outage.sh, check_zone_outage.sh)

documentation

These scripts serve as an example of how to run Krkn-hub scenarios in an automated and reproducible manner. Feel free to open a PR with your suggestions for additional scenarios or use cases.

For a complete overview of our chaos scenarios, please refer to our chaos testing guide, the Krkn GitHub repository, the Krkn-hub GitHub repository, and the Krkn website.

If you'd like to get in touch with the team or just say hello 👋, you can find us on the Kubernetes Slack workspace in the #krkn channel.

And now, release the Krkn!