executor

Documentation

Overview

Package executor defines the executor interface and registry.

Index

• Variables
• func Register(executor Executor)
• type AWSConnectorConfig
• type ConnectorConfig
• type ErrorList
  • func (e *ErrorList) Add(err error)
  • func (e *ErrorList) Addf(format string, args ...interface{})
  • func (e *ErrorList) Join(sep string) string
  • func (e *ErrorList) Len() int
• type Executor
  • func Get(executorType string) (Executor, bool)
• type K8SConnectorConfig
• type Registry
  • func NewRegistry() *Registry
  • func (r *Registry) Get(executorType string) (Executor, bool)
  • func (r *Registry) List() []string
  • func (r *Registry) Register(executor Executor)
• type ReportStateCallback
• type RestoreData
• type Result
• type Spec

Variables

var DefaultRegistry = NewRegistry()

DefaultRegistry is the global executor registry.

Functions

func Register

func Register(executor Executor)

Register adds an executor to the default registry.

Types

type AWSConnectorConfig

type AWSConnectorConfig = awsutil.AWSConnectorConfig

AWSConnectorConfig holds AWS connector settings.

type ConnectorConfig

type ConnectorConfig struct {
	// AWS holds AWS-specific configuration.
	AWS *AWSConnectorConfig
	// K8S holds Kubernetes-specific configuration.
	K8S *K8SConnectorConfig
}

ConnectorConfig holds resolved connector settings.

type ErrorList

type ErrorList struct {
	// contains filtered or unexported fields
}

ErrorList is a thread-safe accumulator for errors. It is useful when multiple goroutines may encounter errors and the caller wants to collect them for later reporting instead of failing fast.

func (*ErrorList) Add

func (e *ErrorList) Add(err error)

Add appends an error to the list. Nil errors are ignored.

func (*ErrorList) Addf

func (e *ErrorList) Addf(format string, args ...interface{})

Addf appends a formatted error to the list.

func (*ErrorList) Join

func (e *ErrorList) Join(sep string) string

Join returns all accumulated error messages joined by the given separator.

func (*ErrorList) Len

func (e *ErrorList) Len() int

Len returns the number of accumulated errors.

type Executor

type Executor interface {
	// Type returns the executor type identifier.
	Type() string

	// Validate validates the executor spec.
	Validate(spec Spec) error

	// Shutdown performs the hibernation operation.
	// Restore data should be saved incrementally via spec.ReportStateCallback.
	// Returns a Result with a summary message on success.
	Shutdown(ctx context.Context, log logr.Logger, spec Spec) (*Result, error)

	// WakeUp performs the restore operation using saved restore data.
	// Returns a Result with a summary message on success.
	WakeUp(ctx context.Context, log logr.Logger, spec Spec, restore RestoreData) (*Result, error)
}

Executor is the interface that all executors must implement.

func Get

func Get(executorType string) (Executor, bool)

Get retrieves an executor from the default registry.

type K8SConnectorConfig

type K8SConnectorConfig = k8sutil.K8SConnectorConfig

K8SConnectorConfig holds Kubernetes connector settings.

type Registry

type Registry struct {
	// contains filtered or unexported fields
}

Registry holds registered executors.

func NewRegistry

func NewRegistry() *Registry

NewRegistry creates a new executor registry.

func (*Registry) Get

func (r *Registry) Get(executorType string) (Executor, bool)

Get retrieves an executor by type.

func (*Registry) List

func (r *Registry) List() []string

List returns all registered executor types.

func (*Registry) Register

func (r *Registry) Register(executor Executor)

Register adds an executor to the registry.

type ReportStateCallback

type ReportStateCallback func(key string, value interface{}) error

ReportStateCallback is a callback for incremental restore data persistence. Executors call this after each successful sub-resource operation to save restore data incrementally, preventing data loss on partial failures. Any data reported through this callback is inherently "live" — the executor just captured it from the actual cloud/cluster API.

Parameters:

key: Resource-specific key (e.g., instanceID, nodeGroupName)
value: Resource state (will be JSON-marshaled by callback implementation)

type RestoreData

type RestoreData struct {
	// Type of the executor that produced this data.
	Type string `json:"type"`

	// Data is a unified map of resource key → resource state (as JSON).
	// Keys are executor-specific:
	// - EC2: instanceID (e.g., "i-1234567890abcdef0")
	// - EKS: nodeGroupName (e.g., "ng-1")
	// - Karpenter: nodePoolName (e.g., "default")
	// - WorkloadScaler: namespace/kind/name (e.g., "team-a/Deployment/api")
	// - RDS: instanceID or clusterID with prefix (e.g., "instance:my-db", "cluster:my-cluster")
	// - Noop: operation ID (e.g., "noop-12345")
	Data map[string]json.RawMessage `json:"data"`

	// IsLive indicates whether data was captured from actual resource state via API (true)
	// or is cached/unknown state (false). When true, hibernator has direct knowledge of the
	// resource state from a fresh API call. When false, the data may be stale or from an
	// incomplete lifecycle. High-quality data (IsLive=true) is preserved over low-quality
	// data (IsLive=false) during save operations.
	IsLive bool `json:"isLive"`
}

RestoreData holds restore metadata produced by Shutdown.

type Result

type Result struct {
	// Message is a human-readable summary of what the executor did.
	// For successful operations, this describes the action taken (e.g., "scaled 3 node groups to zero").
	// For failed operations, the error message is used instead.
	Message string

	// ElapsedMs is the wall-clock execution time in milliseconds.
	// This field is populated by the runner after the executor returns;
	// executor implementations should leave it at zero.
	ElapsedMs int64
}

Result holds the outcome of an executor operation.

type Spec

type Spec struct {
	// TargetName is the name of the target.
	TargetName string
	// TargetType is the type of the target (eks, rds, ec2).
	TargetType string
	// Parameters is the executor-specific configuration.
	Parameters json.RawMessage
	// ConnectorConfig holds resolved connector configuration.
	ConnectorConfig ConnectorConfig
	// ReportStateCallback is an optional callback for incremental persistence.
	// If provided, executors should call this after each successful sub-resource
	// operation to enable partial-success data preservation.
	ReportStateCallback ReportStateCallback
}

Spec holds target execution parameters.