Releases: litmuschaos/litmus
Releases · litmuschaos/litmus
1.3.0-RC1
1.2.2
(Review): Adding OpenEBS NFS provisioner kill experiment (#1248) (#1397) * (feat): Adding OpenEBS NFS provisioner kill experiment Signed-off-by: Raj <[email protected]>
1.2.1
1.2.0
New features and Enhancements
- Addition of Chaos Events (across all litmus components, i.e., operator/runner/experiment job) to indicate experiment lifecycle
- Enhanced ChaosResult with experiment failure reason (step) provided in CR status
- Includes Node Memory Hog experiment to generic/kubernetes suite
- Includes OpenEBS pool disk loss experiment for GKE/AWS
- Adds support for Amazon EKS platform for generic chaos experiments
- Introduces a new chart category based on chaostoolkit with initial pod chaos experiments
- Supports override of default runner properties such as imagePullPolicy & entrypoint/args
- Extends cleanupPolicy enforcement to chaos-runner pods (apart from just the experiment job) with improved reconciliation flow
- Improves experiment chaoslib which now makes use of jobs (replacing daemonsets) to reduce the number of chaos resources (pods) used in an experiment, with chaos injection commands burned into the job templates.
- Adds support for RAMP_UP / RAMP_DOWN periods during the course of a chaos experiment.
- Homogenizes the time units (sec over msec) used across experiments for chaos duration and other parameters.
- Improved e2e suite with Ginkgo based BDD tests for newly added experiments and operator functionality
- Refactors the test-tools repository structure based on tool type
- Introduces an NFS liveness tool to lay foundation for NFS storage chaos experiments
- Adds governance artefacts (Maintainers, Governance) along with the project roadmap and an initial set of public adopters of LitmusChaos
- Adds license dependencies and scan reports obtained via fossa
Major Bug Fixes
- Fixes the hardcoded total chaos/job wait duration in the node-cpu-hog experiment.
- Fixes to verify state of application pods (health check) before proceeding with subsequent iterations of pod-delete chaos
- Adds a unique instance_id/run_id (hash) to names & labels of chaos jobs started by the experiment to aid identification and prevent conflicts upon parallel or repeated runs in a given namespace.
- Fixes execution workflow of chaos experiments when run as a standalone job without orchestration by the chaos operator
Getting Started
Prerequisites to install
- Make sure you have a healthy Kubernetes Cluster.
- Kubernetes 1.11+ is installed
Installation
kubectl apply -f https://litmuschaos.github.io/litmus/litmus-operator-v1.2.0.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
1.2.0-RC1
(chore)roadmap: add issue links to near term roadmap items (#1285) * (chore)roadmap: add issue links to near term roadmap items Signed-off-by: ksatchit <[email protected]> Co-authored-by: Shubham Chaudhary <[email protected]>
1.1.1
1.1.0
New features and Enhancements
- Addition of OpenEBS control-plane sanity, storage pool sanity & storage pool network chaos experiments
- Introduce the Chaos Abort functionality to terminate in-flight chaos experiments at once
- Schema changes to support override of configmap & secret volumes parameters from the ChaosEngine with improved component structure
- Make annotation checks against target applications an optional feature in lieu of infra chaos experiments with higher blast radius (multi-application impact).
- Make the Go Chaos-Runner as the default runner type for experiments
- Improved CRD validation for ChaosEngine CRs based on OpenAPI v3 specification
- Improved logging for experiment lifecycle (chaos-runner) via use of klog framework
- Addition of Kubernetes events for chaos experiment lifecycle
- Enhancement to pass chaos lineage (chaos UUID) across all resources created as part of an experiment
- Strengthened e2e pipelines (generic, openebs chaos validation tests) with BDD tests using Ginkgo
- Updates experiment scaffold (developer) tool to reflect latest experiment schema
- Optimized (minimal) permissions in the chaos-operator RBAC
- Converts the chaos-charts repo into the single source truth wrt chaos manifests (includes experiment RBAC, sample ChaosEngine CR manifests along with individual chaosExperiment CRs & CSV) thereby providing an integrated bundle to aid developers
- Doc updates to embed all chaos snippets from chaos-charts via the embedmd tool
- Adds the Litmus FAQ with questions on general usage & troubleshooting steps
Major Bug Fixes
- Addition of rescue blocks in chaoslib wrappers to fail fast / avoid false positives in chaos experiments
- Fixes container-kill experiment execution on containerd runtime
- Patch chaosresult state with status “running” & verdict “awaited” during experiment execution.
- Fixes crash of exporter (monitor pod) in case of infra experiments with annotationCheck disabled.
- Refactor of experiment manifests to use standard naming conventions & file extensions
Getting Started
Prerequisites to install
- Make sure you have a healthy Kubernetes Cluster.
- Kubernetes 1.11+ is installed
Installation
kubectl apply -f https://litmuschaos.github.io/litmus/litmus-operator-v1.1.0.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
1.1.0-RC2
1.1.0-RC1
(feat): Adding openebs control plane pod delete experiment (#1203) Signed-off-by: Raj <[email protected]>
1.0.1
New features and Enhancements
- Adds ramp time (warm-up period before chaos is injected) support to the generic chaos experiments
- Optimises the chaosServiceAccount permissions needed for current OpenEBS experiments
- Improved negative path/error handling in container-kill experiment
- Support for containerd runtime in OpenEBS chaos experiments
- Improve OpenEBS network loss experiments to use iSCSI session recovery timeout as default loss period with data persistence verification as post-chaos checks