Skip to main content

Chaos Engineering Reference

This section provides comprehensive reference documentation for Harness Chaos Engineering APIs, configurations, and technical specifications.

API Reference​

REST API​

  • Experiments API: Create, manage, and execute chaos experiments
  • Faults API: Access available fault types and configurations
  • Results API: Retrieve experiment results and metrics
  • Infrastructure API: Manage target infrastructure and connections

GraphQL API​

  • Schema Documentation: Complete GraphQL schema reference
  • Query Examples: Common queries and mutations
  • Subscription Events: Real-time experiment updates

Configuration Reference​

Experiment Configuration​

apiVersion: litmuschaos.io/v1alpha1
kind: ChaosExperiment
metadata:
  name: example-experiment
spec:
  definition:
    scope: Cluster
    permissions:
      - apiGroups: [""]
        resources: ["pods"]
        verbs: ["create", "delete", "get", "list"]

Fault Specifications​

  • Infrastructure Faults: CPU, memory, network, and disk fault configurations
  • Application Faults: Service mesh and application-level fault specifications
  • Platform Faults: Kubernetes and cloud provider-specific configurations

Fault Library​

Infrastructure Faults​

  • cpu-hog: CPU stress testing
  • memory-hog: Memory exhaustion testing
  • network-latency: Network delay injection
  • disk-fill: Disk space exhaustion

Kubernetes Faults​

  • pod-delete: Pod termination testing
  • node-drain: Node evacuation testing
  • container-kill: Container failure simulation

Cloud Provider Faults​

  • ec2-terminate: AWS EC2 instance termination
  • vm-stop: Azure VM shutdown
  • gce-stop: GCP Compute Engine instance stop

SDK Reference​

Python SDK​

from harness_chaos import ChaosClient

client = ChaosClient(api_key="your-api-key")
experiment = client.create_experiment(
    name="my-experiment",
    faults=["cpu-hog"],
    targets=["my-service"]
)

Go SDK​

import "github.com/harness/chaos-go-sdk"

client := chaos.NewClient("your-api-key")
experiment, err := client.CreateExperiment(ctx, &chaos.ExperimentRequest{
    Name: "my-experiment",
    Faults: []string{"cpu-hog"},
    Targets: []string{"my-service"},
})

JavaScript SDK​

import { ChaosClient } from '@harness/chaos-js-sdk';

const client = new ChaosClient({ apiKey: 'your-api-key' });
const experiment = await client.createExperiment({
  name: 'my-experiment',
  faults: ['cpu-hog'],
  targets: ['my-service']
});

CLI Reference​

Installation​

# Install Harness Chaos CLI
curl -sSL https://get.harness.io/chaos | bash

Common Commands​

# List available experiments
harness chaos list experiments

# Create a new experiment
harness chaos create experiment --file experiment.yaml

# Run an experiment
harness chaos run experiment my-experiment

# Get experiment results
harness chaos get results my-experiment

Webhook Reference​

Event Types​

  • experiment.started: Experiment execution started
  • experiment.completed: Experiment execution completed
  • experiment.failed: Experiment execution failed
  • fault.injected: Fault injection started
  • fault.recovered: Fault recovery completed

Payload Structure​

{
  "event": "experiment.completed",
  "timestamp": "2023-06-27T10:30:00Z",
  "experiment": {
    "id": "exp-123",
    "name": "my-experiment",
    "status": "completed",
    "result": "passed"
  }
}

Error Codes​

HTTP Status Codes​

  • 200: Success
  • 400: Bad Request - Invalid parameters
  • 401: Unauthorized - Invalid API key
  • 403: Forbidden - Insufficient permissions
  • 404: Not Found - Resource not found
  • 429: Too Many Requests - Rate limit exceeded
  • 500: Internal Server Error

Custom Error Codes​

  • CE001: Experiment validation failed
  • CE002: Target infrastructure unreachable
  • CE003: Insufficient permissions
  • CE004: Fault injection failed
  • CE005: Experiment timeout exceeded

Limits and Quotas​

API Rate Limits​

  • REST API: 1000 requests per minute
  • GraphQL API: 500 queries per minute
  • Webhook Delivery: 100 events per minute

Resource Limits​

  • Concurrent Experiments: 50 per account
  • Experiment Duration: 24 hours maximum
  • Target Resources: 1000 per experiment

Version History​

API Versioning​

  • v1: Current stable version
  • v2: Beta version with enhanced features
  • Legacy: Deprecated versions (v0.x)

Changelog​

See our Release Notes for detailed version history and breaking changes.