Documentation
Guide
Examples
Fault Injection

Fault Injection

🚀 Basic Fault Injection Setup

from rcabench.openapi import (
    ApiClient,
    Configuration,
    DtoSubmitInjectionReq,
    HandlerNode,
    InjectionApi,
)

BASE_URL = "http://localhost:8082"

client = ApiClient(configuration=Configuration(host=BASE_URL))
injector = InjectionApi(api_client=client)
injector.api_v1_injections_post(
    body=DtoSubmitInjectionReq(
      benchmark="ts",
      interval=15,
      pre_duration=15,
      project_name="default",
      specs=[], # Define fault specs here
    )
)

🔧 Fault Injection Scenarios

Scenario 1: Network Delay Injection

Experiment

Test application resilience against network latency issues between microservices. This scenario helps evaluate how your system handles slow network connections and communication delays.

specs=[
    HandlerNode(
        value=17,
        children={
            "17": HandlerNode(
                children={
                    "0": HandlerNode(value=4), # Duration in seconds
                    "1": HandlerNode(value=0), # Namespace (dynamic)
                    "2": HandlerNode(value=1), # NetworkPairIdx (dynamic)
                    "3": HandlerNode(value=1000), # Latency in milliseconds
                    "4": HandlerNode(value=10), # Correlation percentage
                    "5": HandlerNode(value=500), # Jitter in milliseconds
                    "6": HandlerNode(value=1), # Direction (1=to, 2=from, 3=both)
                }
            )
        },
    )
],

Scenario 2: Memory Pressure

Experiment

Evaluate system behavior under high memory consumption conditions. This scenario helps identify memory leaks, inefficient memory usage, and tests memory management strategies.

specs=[
    HandlerNode(
        value=3,
        children={
            "3": HandlerNode(
                children={
                    "0": HandlerNode(value=4), # Duration in seconds
                    "1": HandlerNode(value=0), # Namespace (dynamic)
                    "2": HandlerNode(value=1), # Container Index (dynamic)
                    "3": HandlerNode(value=512), # Memory Size Unit MB
                    "4": HandlerNode(value=1), # Memory Stress Threads
                }
            )
        },
    )
],

Scenario 3: Pod Failure

Experiment

Test system fault tolerance and recovery mechanisms when pods unexpectedly terminate. This scenario validates auto-scaling, service mesh resilience, and application restart policies.

specs=[
    HandlerNode(
        value=3,
        children={
            "3": HandlerNode(
                children={
                    "0": HandlerNode(value=4), # Duration in seconds
                    "1": HandlerNode(value=0), # Namespace (dynamic)
                    "2": HandlerNode(value=1), # App Index (dynamic)
                }
            )
        },
    )
],

📚 Reference

For more fault types and parameters, see: