chaosmonkey

README

Chaos Monkey randomly terminates virtual machine instances and containers that run inside of your production environment. Exposing engineers to failures more frequently incentivizes them to build resilient services.

See the documentation for info on how to use Chaos Monkey.

Chaos Monkey is an example of a tool that follows the Principles of Chaos Engineering.

Requirements

This version of Chaos Monkey is fully integrated with Spinnaker, the continuous delivery platform that we use at Netflix. You must be managing your apps with Spinnaker to use Chaos Monkey to terminate instances.

Chaos Monkey should work with any backend that Spinnaker supports (AWS, Google Compute Engine, Azure, Kubernetes, Cloud Foundry). It has been tested with AWS, GCE, and Kubernetes.

Install locally

To install the Chaos Monkey binary on your local machine:

go get github.com/netflix/chaosmonkey/cmd/chaosmonkey

How to deploy

See the docs for instructions on how to configure and deploy Chaos Monkey.

Support

Simian Army Google group.

Documentation

Overview

Package chaosmonkey contains our domain models

Index

• type AppConfig
  • func NewAppConfig(exceptions []Exception) AppConfig
• type AppConfigGetter
• type Checker
• type Decryptor
• type Env
• type ErrViolatesMinTime
  • func (e ErrViolatesMinTime) Error() string
• type ErrorCounter
• type Exception
  • func (ex Exception) Matches(account, stack, detail, region string) bool
• type Group
  • func (g Group) String() string
• type Instance
• type Outage
• type Termination
• type Terminator
• type Tracker

Types

type AppConfig

type AppConfig struct {
	Enabled                        bool
	RegionsAreIndependent          bool
	MeanTimeBetweenKillsInWorkDays int
	MinTimeBetweenKillsInWorkDays  int
	Grouping                       Group
	Exceptions                     []Exception
	Whitelist                      *[]Exception
}

AppConfig contains app-specific configuration parameters for Chaos Monkey

func NewAppConfig

func NewAppConfig(exceptions []Exception) AppConfig

NewAppConfig constructs a new app configuration with reasonable defaults with specified accounts enabled/disabled

type AppConfigGetter

type AppConfigGetter interface {
	// Get returns the App config info by app name
	Get(app string) (*AppConfig, error)
}

AppConfigGetter retrieves App configuration info

type Checker

type Checker interface {
	// Check checks if a termination is permitted and, if so, records the
	// termination time on the server.
	// The endHour (hour time when Chaos Monkey stops killing) is in the
	// time zone specified by loc.
	Check(term Termination, appCfg AppConfig, endHour int, loc *time.Location) error
}

Checker checks to see if a termination is permitted given min time between terminations

if the termination is permitted, returns (true, nil) otherwise, returns false with an error

Returns ErrViolatesMinTime if violates min time between terminations

Note that this call may change the state of the server: if the checker returns true, the termination will be recorded.

type Decryptor

type Decryptor interface {
	Decrypt(ciphertext string) (string, error)
}

Decryptor decrypts encrypted text. It is used for decrypting sensitive credentials that are stored encrypted

type Env

type Env interface {
	// InTest returns true if Chaos Monkey is running in a test environment
	InTest() bool
}

Env provides information about the environment that Chaos Monkey has been deployed to.

type ErrViolatesMinTime

type ErrViolatesMinTime struct {
	InstanceID string         // the most recent terminated instance id
	KilledAt   time.Time      // the time that the most recent instance was terminated
	Loc        *time.Location // local time zone location
}

ErrViolatesMinTime represents an error when trying to record a termination that violates the min time between terminations for that particular app

func (ErrViolatesMinTime) Error

func (e ErrViolatesMinTime) Error() string

type ErrorCounter

type ErrorCounter interface {
	Increment() error
}

ErrorCounter counts when errors occur.

type Exception

type Exception struct {
	Account string
	Stack   string
	Detail  string
	Region  string
}

Exception describes clusters that have been opted out of chaos monkey If one of the members is a "*", it matches everything. That is the only wildcard value For example, this will opt-out all of the cluters in the test account: Exception{ Account:"test", Stack:"*", Cluster:"*", Region: "*"}

func (Exception) Matches

func (ex Exception) Matches(account, stack, detail, region string) bool

Matches returns true if an exception matches an ASG

type Group

type Group int

Group describes what Chaos Monkey considers a group of instances Chaos Monkey will randomly kill an instance from each group. The group generally maps onto what the service owner considers a "cluster", which is different from Spinnaker's notion of a cluster.

const (
	// App grouping: Chaos Monkey kills one instance per app per day
	App Group = iota
	// Stack grouping: Chaos Monkey kills one instance per stack per day
	Stack
	// Cluster grouping: Chaos Monkey kills one instance per cluster per day
	Cluster
)

func (Group) String

func (g Group) String() string

String returns a string representation for a Group

type Instance

type Instance interface {
	// AppName is the name of the Netflix app
	AppName() string

	// AccountName is the name of the account the instance is running in (e.g., prod, test)
	AccountName() string

	// RegionName is the name of the AWS region (e.g., us-east-1
	RegionName() string

	// StackName returns the "stack" part of app-stack-detail in cluster names
	StackName() string

	// ClusterName is the full cluster name: app-stack-detail
	ClusterName() string

	// ASGName is the name of the ASG associated with the instance
	ASGName() string

	// ID is the instance ID, e.g. i-dbcba24c
	ID() string

	// CloudProvider returns the cloud provider (e.g., "aws")
	CloudProvider() string
}

Instance contains naming info about an instance

type Outage

type Outage interface {
	// Outage returns true if there is an ongoing outage
	Outage() (bool, error)
}

Outage provides an interface for checking if there is currently an outage This provides a mechanism to check if there's an ongoing outage, since Chaos Monkey doesn't run during outages

type Termination

type Termination struct {
	Instance Instance  // The instance that will be terminated
	Time     time.Time // Termination time
	Leashed  bool      // If true, track the termination but do not execute it
}

Termination contains information about an instance termination.

type Terminator

type Terminator interface {
	// Kill terminates a running instance
	Execute(trm Termination) error
}

Terminator provides an interface for killing instances

type Tracker

type Tracker interface {
	// Track pushes a termination event to the tracking system
	Track(t Termination) error
}

Tracker records termination events an a tracking system such as Chronos