New Features & Enhancements

• Introduces category for VMWare chaos with VM power-off experiment (supported for vCenter 6.x)

• Adds chaos experiments for simulating DNS errors (inability to resolve hosts) and redirection to incorrect/faulty services (using a spoof map that can redirect specific requests)

• Makes the chaos annotationCheck against applications “false” by default, making it simpler for users to get started with chaos without any instrumentation step for the application targets.

• Updates the CRD version to v1, the min. supported Kubernetes version moved to 1.15

• Enhances the disk fill experiment with a tunable to specify write block size for quicker capacity use and fs aligned writes.

• Supports label-based selection of node targets for (node-level) chaos injection.

• Adds chaos abort routines for AWS chaos experiments

• Adds the ability to target EBS volumes by tag, with a sequential and parallel injection of chaos, with support for both simple as well as EKS persistent volumes.

• Places non-litmus core images (dependencies such as argo, MongoDB for portal driven chaos) into litmuschaos image registry, while maintaining image names and release tags to simplify the user experience for those who need to set up local mirrors or are in air-gapped environments

• Adds support for Openshift Route in the litmus helm charts

• Refactors and optimizes chaos libraries for code reuse and simplified flow. Updates the litmus-sdk to generate refactored experiment templates

• Adds GitHub actions based workflow/pipeline for node-level chaos experiments in e2e suite

Major Bug Fixes

• Fixes the inability to define certain attributes within the ChaosEngines, for which the OpenAPI validation was missing (due to migration of CRD version to v1) using the “preserve-unknown-fields” option. Also adds the validations for a number of properties/attributes.
  Fixes a panic encountered in the chaos-runner upon the inability to access the ChaosEngine resource

• Fixes the node restart experiment to perform the right verification checks on helper pods executing the chaos
  Fixes behavior where helper pods that complete quickly (run for short durations) are treated as failed by verifying for “succeeded” state.

• Removes ambiguity in filtering/accessing helper pods by assigning standard label format

• Fixes an erroneous decision in pod-cpu & memory hog experiments which considered a non-zero response (137) upon chaos process kill (SIGKILL) as failure to revert/rollback

• Adds a check to verify the status of application target containers before attempting an exec operation to perform the desired chaos action

• Fixes the ec2-terminate-by-tag experiment to consider only the running instances for stop/termination

• Adds the missing PORTAL_ENDPOINT environment to facilitate namespaced mode of execution of the litmus-portal

Installation

kubectl apply -f https://litmuschaos.github.io/litmus/litmus-operator-v1.13.5.yaml

Verify your installation

• Verify if the chaos operator is running
  kubectl get pods -n litmus

• Verify if chaos CRDs are installed
  kubectl get crds | grep chaos

For more details refer to the documentation at Docs