tasks/service/history/queues/scheduler.go

// The MIT License
//
// Copyright (c) 2020 Temporal Technologies Inc.  All rights reserved.
//
// Copyright (c) 2020 Uber Technologies, Inc.
//
// Permission is hereby granted, free of charge, to any person obtaining a copy
// of this software and associated documentation files (the "Software"), to deal
// in the Software without restriction, including without limitation the rights
// to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
// copies of the Software, and to permit persons to whom the Software is
// furnished to do so, subject to the following conditions:
//
// The above copyright notice and this permission notice shall be included in
// all copies or substantial portions of the Software.
//
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
// THE SOFTWARE.

//go:generate mockgen -copyright_file ../../../LICENSE -package $GOPACKAGE -source $GOFILE -destination scheduler_mock.go

package queues

import (
	"go.temporal.io/server/common/clock"
	"go.temporal.io/server/common/dynamicconfig"
	"go.temporal.io/server/common/log"
	"go.temporal.io/server/common/log/tag"
	"go.temporal.io/server/common/metrics"
	"go.temporal.io/server/common/namespace"
	"go.temporal.io/server/common/quotas"
	"go.temporal.io/server/common/tasks"
	"go.temporal.io/server/service/history/configs"
)

const (
	// This is the task channel buffer size between
	// weighted round robin scheduler and the actual
	// worker pool (fifo processor).
	prioritySchedulerProcessorQueueSize = 10

	taskSchedulerToken = 1
)

type (
	// Scheduler is the component for scheduling and processing
	// task executables, it's based on the common/tasks.Scheduler
	// interface and provide the additional information of how
	// tasks are grouped during scheduling.
	Scheduler interface {
		Start()
		Stop()
		Submit(Executable)
		TrySubmit(Executable) bool

		TaskChannelKeyFn() TaskChannelKeyFn
	}

	TaskChannelKey struct {
		NamespaceID string
		Priority    tasks.Priority
	}

	TaskChannelKeyFn = tasks.TaskChannelKeyFn[Executable, TaskChannelKey]
	ChannelWeightFn  = tasks.ChannelWeightFn[TaskChannelKey]

	SchedulerOptions struct {
		WorkerCount             dynamicconfig.IntPropertyFn
		ActiveNamespaceWeights  dynamicconfig.MapPropertyFnWithNamespaceFilter
		StandbyNamespaceWeights dynamicconfig.MapPropertyFnWithNamespaceFilter
	}

	RateLimitedSchedulerOptions struct {
		EnableShadowMode dynamicconfig.BoolPropertyFn
		StartupDelay     dynamicconfig.DurationPropertyFn
	}

	schedulerImpl struct {
		tasks.Scheduler[Executable]
		namespaceRegistry namespace.Registry

		taskChannelKeyFn      TaskChannelKeyFn
		channelWeightFn       ChannelWeightFn
		channelWeightUpdateCh chan struct{}
	}

	rateLimitedSchedulerImpl struct {
		tasks.Scheduler[Executable]

		baseScheduler Scheduler
	}
)

func NewScheduler(
	currentClusterName string,
	options SchedulerOptions,
	namespaceRegistry namespace.Registry,
	logger log.Logger,
) Scheduler {
	var scheduler tasks.Scheduler[Executable]

	taskChannelKeyFn := func(e Executable) TaskChannelKey {
		return TaskChannelKey{
			NamespaceID: e.GetNamespaceID(),
			Priority:    e.GetPriority(),
		}
	}
	channelWeightFn := func(key TaskChannelKey) int {
		namespaceWeights := options.ActiveNamespaceWeights
		namespaceName := namespace.EmptyName

		ns, err := namespaceRegistry.GetNamespaceByID(namespace.ID(key.NamespaceID))
		if err == nil {
			namespaceName = ns.Name()
			if !ns.ActiveInCluster(currentClusterName) {
				namespaceWeights = options.StandbyNamespaceWeights
			}
		} else {
			// if namespace not found, treat it as active namespace and
			// use default active namespace weight
			logger.Warn("Unable to find namespace, using active namespace task channel weight",
				tag.WorkflowNamespaceID(key.NamespaceID),
				tag.Error(err),
			)
		}

		return configs.ConvertDynamicConfigValueToWeights(
			namespaceWeights(namespaceName.String()),
			logger,
		)[key.Priority]
	}
	channelWeightUpdateCh := make(chan struct{}, 1)
	fifoSchedulerOptions := &tasks.FIFOSchedulerOptions{
		QueueSize:   prioritySchedulerProcessorQueueSize,
		WorkerCount: options.WorkerCount,
	}

	scheduler = tasks.NewInterleavedWeightedRoundRobinScheduler(
		tasks.InterleavedWeightedRoundRobinSchedulerOptions[Executable, TaskChannelKey]{
			TaskChannelKeyFn:      taskChannelKeyFn,
			ChannelWeightFn:       channelWeightFn,
			ChannelWeightUpdateCh: channelWeightUpdateCh,
		},
		tasks.Scheduler[Executable](tasks.NewFIFOScheduler[Executable](
			fifoSchedulerOptions,
			logger,
		)),
		logger,
	)

	return &schedulerImpl{
		Scheduler:             scheduler,
		namespaceRegistry:     namespaceRegistry,
		taskChannelKeyFn:      taskChannelKeyFn,
		channelWeightFn:       channelWeightFn,
		channelWeightUpdateCh: channelWeightUpdateCh,
	}
}

func (s *schedulerImpl) Start() {
	if s.channelWeightUpdateCh != nil {
		s.namespaceRegistry.RegisterStateChangeCallback(s, func(ns *namespace.Namespace, deletedFromDb bool) {
			select {
			case s.channelWeightUpdateCh <- struct{}{}:
			default:
			}
		})
	}
	s.Scheduler.Start()
}

func (s *schedulerImpl) Stop() {
	if s.channelWeightUpdateCh != nil {
		s.namespaceRegistry.UnregisterStateChangeCallback(s)

		// note we can't close the channelWeightUpdateCh here
		// as callback may still be triggered even after unregister returns
		// due to race condition
		//
		// channelWeightFn is only not nil when using host level scheduler
		// so Stop is only called when host is shutting down, and we don't need
		// to worry about open channels
	}
	s.Scheduler.Stop()
}

func (s *schedulerImpl) TaskChannelKeyFn() TaskChannelKeyFn {
	return s.taskChannelKeyFn
}

func NewRateLimitedScheduler(
	baseScheduler Scheduler,
	options RateLimitedSchedulerOptions,
	currentClusterName string,
	namespaceRegistry namespace.Registry,
	rateLimiter SchedulerRateLimiter,
	timeSource clock.TimeSource,
	logger log.Logger,
	metricsHandler metrics.Handler,
) Scheduler {
	if delay := options.StartupDelay(); delay > 0 {
		delayedRateLimiter, err := quotas.NewDelayedRequestRateLimiter(
			rateLimiter,
			delay,
			timeSource,
		)
		if err != nil {
			logger.Error("Failed to create delayed rate limited scheduler", tag.Error(err))
			return baseScheduler
		}

		rateLimiter = delayedRateLimiter
	}

	taskQuotaRequestFn := func(e Executable) quotas.Request {
		namespaceName, err := namespaceRegistry.GetNamespaceName(namespace.ID(e.GetNamespaceID()))
		if err != nil {
			namespaceName = namespace.EmptyName
		}
		return quotas.NewRequest("", taskSchedulerToken, namespaceName.String(), tasks.PriorityName[e.GetPriority()], 0, "")
	}
	taskMetricsTagsFn := func(e Executable) []metrics.Tag {
		return append(EstimateTaskMetricTag(e, namespaceRegistry, currentClusterName), metrics.TaskPriorityTag(e.GetPriority().String()))
	}

	rateLimitedScheduler := tasks.NewRateLimitedScheduler[Executable](
		baseScheduler,
		rateLimiter,
		timeSource,
		taskQuotaRequestFn,
		taskMetricsTagsFn,
		tasks.RateLimitedSchedulerOptions{
			EnableShadowMode: options.EnableShadowMode(),
		},
		logger,
		metricsHandler,
	)

	return &rateLimitedSchedulerImpl{
		Scheduler:     rateLimitedScheduler,
		baseScheduler: baseScheduler,
	}
}

func (s *rateLimitedSchedulerImpl) Start() {
	s.baseScheduler.Start()
}

func (s *rateLimitedSchedulerImpl) Stop() {
	s.baseScheduler.Stop()
}

func (s *rateLimitedSchedulerImpl) TaskChannelKeyFn() TaskChannelKeyFn {
	return s.baseScheduler.TaskChannelKeyFn()
}