config

package
Version: v0.10.3 Latest Latest
Warning

This package is not in the latest version of its module.

Go to latest
Published: Apr 8, 2021 License: Apache-2.0 Imports: 7 Imported by: 0

Documentation

Overview

Package config contains the core configuration for FlytePropeller. This configuration can be added under the “propeller“ section.

Example config:

----------------

propeller:
   rawoutput-prefix: s3://my-container/test/
   metadata-prefix: metadata/propeller/sandbox
   workers: 4
   workflow-reeval-duration: 10s
   downstream-eval-duration: 5s
   limit-namespace: "all"
   prof-port: 11254
   metrics-prefix: flyte
   enable-admin-launcher: true
   max-ttl-hours: 1
   gc-interval: 500m
   queue:
     type: batch
     queue:
       type: bucket
       rate: 20
       capacity: 100
     sub-queue:
       type: bucket
       rate: 100
       capacity: 1000
   # This config assumes using `make start` in flytesnacks repo to startup a DinD k3s container
   kube-config: "$HOME/kubeconfig/k3s/k3s.yaml"
   publish-k8s-events: true
   workflowStore:
     policy: "ResourceVersionCache"

Index

Constants

This section is empty.

Variables

This section is empty.

Functions

func MustRegisterSubSection added in v0.1.13

func MustRegisterSubSection(subSectionKey string, section config.Config) config.Section

MustRegisterSubSection can be used to configure any subsections the the propeller configuration

Types

type CompositeQueueConfig

type CompositeQueueConfig struct {
	Type             CompositeQueueType `json:"type" pflag:",Type of composite queue to use for the WorkQueue"`
	Queue            WorkqueueConfig    `json:"queue,omitempty" pflag:",Workflow workqueue configuration, affects the way the work is consumed from the queue."`
	Sub              WorkqueueConfig    `` /* 130-byte string literal not displayed */
	BatchingInterval config.Duration    `json:"batching-interval" pflag:",Duration for which downstream updates are buffered"`
	BatchSize        int                `` /* 135-byte string literal not displayed */
}

CompositeQueueConfig contains configuration for the controller queue and the downstream resource queue

type CompositeQueueType

type CompositeQueueType = string
const (
	CompositeQueueSimple CompositeQueueType = "simple"
	CompositeQueueBatch  CompositeQueueType = "batch"
)

type Config

type Config struct {
	KubeConfigPath         string               `json:"kube-config" pflag:",Path to kubernetes client config file."`
	MasterURL              string               `json:"master"`
	Workers                int                  `json:"workers" pflag:"2,Number of threads to process workflows"`
	WorkflowReEval         config.Duration      `json:"workflow-reeval-duration" pflag:"\"30s\",Frequency of re-evaluating workflows"`
	DownstreamEval         config.Duration      `json:"downstream-eval-duration" pflag:"\"60s\",Frequency of re-evaluating downstream tasks"`
	LimitNamespace         string               `json:"limit-namespace" pflag:"\"all\",Namespaces to watch for this propeller"`
	ProfilerPort           config.Port          `json:"prof-port" pflag:"\"10254\",Profiler port"`
	MetadataPrefix         string               `` /* 244-byte string literal not displayed */
	DefaultRawOutputPrefix string               `` /* 138-byte string literal not displayed */
	Queue                  CompositeQueueConfig `json:"queue,omitempty" pflag:",Workflow workqueue configuration, affects the way the work is consumed from the queue."`
	MetricsPrefix          string               `json:"metrics-prefix" pflag:"\"flyte:\",An optional prefix for all published metrics."`
	EnableAdminLauncher    bool                 `json:"enable-admin-launcher" pflag:"false, Enable remote Workflow launcher to Admin"`
	MaxWorkflowRetries     int                  `json:"max-workflow-retries" pflag:"50,Maximum number of retries per workflow"`
	MaxTTLInHours          int                  `json:"max-ttl-hours" pflag:"23,Maximum number of hours a completed workflow should be retained. Number between 1-23 hours"`
	GCInterval             config.Duration      `json:"gc-interval" pflag:"\"30m\",Run periodic GC every 30 minutes"`
	LeaderElection         LeaderElectionConfig `json:"leader-election,omitempty" pflag:",Config for leader election."`
	PublishK8sEvents       bool                 `json:"publish-k8s-events" pflag:",Enable events publishing to K8s events API."`
	MaxDatasetSizeBytes    int64                `json:"max-output-size-bytes" pflag:",Maximum size of outputs per task"`
	KubeConfig             KubeClientConfig     `json:"kube-client-config" pflag:",Configuration to control the Kubernetes client"`
	NodeConfig             NodeConfig           `json:"node-config,omitempty" pflag:",config for a workflow node"`
	MaxStreakLength        int                  `` /* 152-byte string literal not displayed */
}

Config that uses the flytestdlib Config module to generate commandline and load config files. This configuration is the base configuration to start propeller NOTE: when adding new fields, do not mark them as "omitempty" if it's desirable to read the value from env variables.

func GetConfig

func GetConfig() *Config

GetConfig extracts the Configuration from the global config module in flytestdlib and returns the corresponding type-casted object.

func (Config) GetPFlagSet

func (cfg Config) GetPFlagSet(prefix string) *pflag.FlagSet

GetPFlagSet will return strongly types pflags for all fields in Config and its nested types. The format of the flags is json-name.json-sub-name... etc.

type DefaultDeadlines added in v0.1.17

type DefaultDeadlines struct {
	DefaultNodeExecutionDeadline  config.Duration `json:"node-execution-deadline" pflag:",Default value of node execution timeout"`
	DefaultNodeActiveDeadline     config.Duration `json:"node-active-deadline" pflag:",Default value of node timeout"`
	DefaultWorkflowActiveDeadline config.Duration `json:"workflow-active-deadline" pflag:",Default value of workflow timeout"`
}

DefaultDeadlines contains default values for timeouts

type KubeClientConfig added in v0.1.8

type KubeClientConfig struct {
	// QPS indicates the maximum QPS to the master from this client.
	// If it's zero, the created RESTClient will use DefaultQPS: 5
	QPS float32 `json:"qps" pflag:"-,Max QPS to the master for requests to KubeAPI. 0 defaults to 5."`
	// Maximum burst for throttle.
	// If it's zero, the created RESTClient will use DefaultBurst: 10.
	Burst int `json:"burst" pflag:",Max burst rate for throttle. 0 defaults to 10"`
	// The maximum length of time to wait before giving up on a server request. A value of zero means no timeout.
	Timeout config.Duration `json:"timeout" pflag:",Max duration allowed for every request to KubeAPI before giving up. 0 implies no timeout."`
}

KubeClientConfig contains the configuration used by flytepropeller to configure its internal Kubernetes Client.

type LeaderElectionConfig

type LeaderElectionConfig struct {
	// Enable or disable leader election.
	Enabled bool `json:"enabled" pflag:",Enables/Disables leader election."`

	// Determines the name of the configmap that leader election will use for holding the leader lock.
	LockConfigMap types.NamespacedName `json:"lock-config-map" pflag:",ConfigMap namespace/name to use for resource lock."`

	// Duration that non-leader candidates will wait to force acquire leadership. This is measured against time of last
	// observed ack
	LeaseDuration config.Duration `` /* 157-byte string literal not displayed */

	// RenewDeadline is the duration that the acting master will retry refreshing leadership before giving up.
	RenewDeadline config.Duration `json:"renew-deadline" pflag:",Duration that the acting master will retry refreshing leadership before giving up."`

	// RetryPeriod is the duration the LeaderElector clients should wait between tries of actions.
	RetryPeriod config.Duration `json:"retry-period" pflag:",Duration the LeaderElector clients should wait between tries of actions."`
}

LeaderElectionConfig Contains leader election configuration.

type NodeConfig added in v0.7.0

type NodeConfig struct {
	DefaultDeadlines               DefaultDeadlines `json:"default-deadlines,omitempty" pflag:",Default value for timeouts"`
	MaxNodeRetriesOnSystemFailures int64            `json:"max-node-retries-system-failures" pflag:"2,Maximum number of retries per node for node failure due to infra issues"`
	InterruptibleFailureThreshold  int64            `json:"interruptible-failure-threshold" pflag:"1,number of failures for a node to be still considered interruptible'"`
}

NodeConfig contains configuration that is useful for every node execution

type WorkqueueConfig

type WorkqueueConfig struct {
	// Refer to https://github.com/kubernetes/client-go/tree/master/util/workqueue
	Type      WorkqueueType   `json:"type" pflag:",Type of RateLimiter to use for the WorkQueue"`
	BaseDelay config.Duration `json:"base-delay" pflag:",base backoff delay for failure"`
	MaxDelay  config.Duration `json:"max-delay" pflag:",Max backoff delay for failure"`
	Rate      int64           `json:"rate" pflag:",Bucket Refill rate per second"`
	Capacity  int             `json:"capacity" pflag:",Bucket capacity as number of items"`
}

WorkqueueConfig has the configuration to configure a workqueue. We may want to generalize this in a package like k8sutils

type WorkqueueType

type WorkqueueType = string
const (
	WorkqueueTypeDefault                       WorkqueueType = "default"
	WorkqueueTypeBucketRateLimiter             WorkqueueType = "bucket"
	WorkqueueTypeExponentialFailureRateLimiter WorkqueueType = "expfailure"
	WorkqueueTypeMaxOfRateLimiter              WorkqueueType = "maxof"
)

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL