tasks/service/worker/scanner/executions/scavenger.go

package executions

import (
	"context"
	"sync"
	"sync/atomic"
	"time"

	"go.temporal.io/server/api/adminservice/v1"
	"go.temporal.io/server/api/historyservice/v1"
	"go.temporal.io/server/common"
	"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/persistence"
	"go.temporal.io/server/common/quotas"
	"go.temporal.io/server/service/worker/scanner/executor"
)

const (
	executorPollInterval     = time.Minute
	executorMaxDeferredTasks = 50000
)

type (
	// Scavenger is the type that holds the state for executions scavenger daemon
	Scavenger struct {
		status           int32
		numHistoryShards int32
		activityContext  context.Context

		executionManager              persistence.ExecutionManager
		registry                      namespace.Registry
		historyClient                 historyservice.HistoryServiceClient
		adminClient                   adminservice.AdminServiceClient
		executor                      executor.Executor
		rateLimiter                   quotas.RateLimiter
		perShardQPS                   dynamicconfig.IntPropertyFn
		executionDataDurationBuffer   dynamicconfig.DurationPropertyFn
		enableHistoryEventIDValidator dynamicconfig.BoolPropertyFn
		metricsHandler                metrics.Handler
		logger                        log.Logger

		stopC  chan struct{}
		stopWG sync.WaitGroup
	}
)

// NewScavenger returns an instance of executions scavenger daemon
// The Scavenger can be started by calling the Start() method on the
// returned object. Calling the Start() method will result in one
// complete iteration over all of the open workflow executions in the system. For
// each executions, will attempt to validate the workflow execution and emit metrics/logs on validation failures.
//
// The scavenger will retry on all persistence errors infinitely and will only stop under
// two conditions
//   - either all executions are processed successfully (or)
//   - Stop() method is called to stop the scavenger
func NewScavenger(
	activityContext context.Context,
	numHistoryShards int32,
	perHostQPS dynamicconfig.IntPropertyFn,
	perShardQPS dynamicconfig.IntPropertyFn,
	executionDataDurationBuffer dynamicconfig.DurationPropertyFn,
	executionTaskWorker dynamicconfig.IntPropertyFn,
	enableHistoryEventIDValidator dynamicconfig.BoolPropertyFn,
	executionManager persistence.ExecutionManager,
	registry namespace.Registry,
	historyClient historyservice.HistoryServiceClient,
	adminClient adminservice.AdminServiceClient,
	metricsHandler metrics.Handler,
	logger log.Logger,
) *Scavenger {
	return &Scavenger{
		activityContext:  activityContext,
		numHistoryShards: numHistoryShards,
		executionManager: executionManager,
		registry:         registry,
		historyClient:    historyClient,
		adminClient:      adminClient,
		executor: executor.NewFixedSizePoolExecutor(
			executionTaskWorker(),
			executorMaxDeferredTasks,
			metricsHandler,
			metrics.ExecutionsScavengerScope,
		),
		rateLimiter: quotas.NewDefaultOutgoingRateLimiter(
			func() float64 { return float64(perHostQPS()) },
		),
		perShardQPS:                   perShardQPS,
		executionDataDurationBuffer:   executionDataDurationBuffer,
		enableHistoryEventIDValidator: enableHistoryEventIDValidator,
		metricsHandler:                metricsHandler.WithTags(metrics.OperationTag(metrics.ExecutionsScavengerScope)),
		logger:                        logger,

		stopC: make(chan struct{}),
	}
}

// Start starts the scavenger
func (s *Scavenger) Start() {
	if !atomic.CompareAndSwapInt32(
		&s.status,
		common.DaemonStatusInitialized,
		common.DaemonStatusStarted,
	) {
		return
	}
	s.logger.Info("Executions scavenger starting")
	s.stopWG.Add(1)
	s.executor.Start()
	go s.run()
	metrics.StartedCount.With(s.metricsHandler).Record(1)
	s.logger.Info("Executions scavenger started")
}

// Stop stops the scavenger
func (s *Scavenger) Stop() {
	if !atomic.CompareAndSwapInt32(
		&s.status,
		common.DaemonStatusStarted,
		common.DaemonStatusStopped,
	) {
		return
	}
	metrics.StoppedCount.With(s.metricsHandler).Record(1)
	s.logger.Info("Executions scavenger stopping")
	close(s.stopC)
	s.executor.Stop()
	s.stopWG.Wait()
	s.logger.Info("Executions scavenger stopped")
}

// Alive returns true if the scavenger is still running
func (s *Scavenger) Alive() bool {
	return atomic.LoadInt32(&s.status) == common.DaemonStatusStarted
}

// run does a single run over all executions and validates them
func (s *Scavenger) run() {
	defer func() {
		go s.Stop()
		s.stopWG.Done()
	}()

	for shardID := int32(1); shardID <= s.numHistoryShards; shardID++ {
		submitted := s.executor.Submit(newTask(
			s.activityContext,
			shardID,
			s.executionManager,
			s.registry,
			s.historyClient,
			s.adminClient,
			s.metricsHandler,
			s.logger,
			s,
			quotas.NewMultiRateLimiter([]quotas.RateLimiter{
				quotas.NewDefaultOutgoingRateLimiter(
					func() float64 { return float64(s.perShardQPS()) },
				),
				s.rateLimiter,
			}),
			s.executionDataDurationBuffer,
			s.enableHistoryEventIDValidator,
		))
		if !submitted {
			s.logger.Error("unable to submit task to executor", tag.ShardID(shardID))
		}
	}

	s.awaitExecutor()
}

func (s *Scavenger) awaitExecutor() {
	// gauge value persists, so we want to reset it to 0
	defer metrics.ExecutionsOutstandingCount.With(s.metricsHandler).Record(float64(0))

	outstanding := s.executor.TaskCount()
	for outstanding > 0 {
		timer := time.NewTimer(executorPollInterval)
		select {
		case <-timer.C:
			outstanding = s.executor.TaskCount()
			metrics.ExecutionsOutstandingCount.With(s.metricsHandler).Record(float64(outstanding))
		case <-s.stopC:
			timer.Stop()
			return
		}
	}
}