v0.9.4

running

.travis.yml

@@ -5,6 +5,7 @@ git:
 go: [1.13.x, 1.14.x]
 script:
   - go test -race -v -coverprofile=coverage.txt -covermode=atomic ./...
+  - go test -run=XXX -bench=. -loglevel=debug ./...
 services:
   - redis-server
 after_success:

.travis/benchcmp.sh

@@ -3,13 +3,16 @@ if [ "${TRAVIS_PULL_REQUEST_BRANCH:-$TRAVIS_BRANCH}" != "master" ]; then
     cd ${TRAVIS_BUILD_DIR}/.. && \
     git clone ${REMOTE_URL} "${TRAVIS_REPO_SLUG}-bench" && \
     cd "${TRAVIS_REPO_SLUG}-bench" && \
+
     # Benchmark master
     git checkout master && \
     go test -run=XXX -bench=. ./... > master.txt && \
+
     # Benchmark feature branch
     git checkout ${TRAVIS_COMMIT} && \
     go test -run=XXX -bench=. ./... > feature.txt && \
-    go get -u golang.org/x/tools/cmd/benchcmp && \
+
     # compare two benchmarks
+    go get -u golang.org/x/tools/cmd/benchcmp && \
     benchcmp master.txt feature.txt;
 fi

CHANGELOG.md

@@ -7,6 +7,25 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## [Unreleased]
 
+## [0.9.4] - 2020-06-13
+
+### Fixed
+
+- Fixes issue of same tasks processed by more than one worker (https://github.com/hibiken/asynq/issues/90).
+
+## [0.9.3] - 2020-06-12
+
+### Fixed
+
+- Fixes the JSON number overflow issue (https://github.com/hibiken/asynq/issues/166).
+
+
+## [0.9.2] - 2020-06-08
+
+### Added
+
+- The `pause` and `unpause` commands were added to the CLI. See README for the CLI for details.
+
 ## [0.9.1] - 2020-05-29
 
 ### Added

README.md

@@ -40,6 +40,7 @@ A system can consist of multiple worker servers and brokers, giving way to high
 - De-duplication of tasks using [unique option](https://github.com/hibiken/asynq/wiki/Unique-Tasks)
 - Allow [timeout and deadline per task](https://github.com/hibiken/asynq/wiki/Task-Timeout-and-Cancelation)
 - [Flexible handler interface with support for middlewares](https://github.com/hibiken/asynq/wiki/Handler-Deep-Dive)
+- [Ability to pause queue](/tools/asynq/README.md#pause) to stop processing tasks from the queue
 - [Support Redis Sentinels](https://github.com/hibiken/asynq/wiki/Automatic-Failover) for HA
 - [CLI](#command-line-tool) to inspect and remote-control queues and tasks
 

benchmark_test.go

@@ -7,7 +7,6 @@ package asynq
 import (
 	"context"
 	"fmt"
-	"math/rand"
 	"sync"
 	"testing"
 	"time"
@@ -29,6 +28,7 @@ func BenchmarkEndToEndSimple(b *testing.B) {
 			RetryDelayFunc: func(n int, err error, t *Task) time.Duration {
 				return time.Second
 			},
+			LogLevel: testLogLevel,
 		})
 		// Create a bunch of tasks
 		for i := 0; i < count; i++ {
@@ -60,7 +60,6 @@ func BenchmarkEndToEnd(b *testing.B) {
 	const count = 100000
 	for n := 0; n < b.N; n++ {
 		b.StopTimer() // begin setup
-		rand.Seed(time.Now().UnixNano())
 		setup(b)
 		redis := &RedisClientOpt{
 			Addr: redisAddr,
@@ -72,6 +71,7 @@ func BenchmarkEndToEnd(b *testing.B) {
 			RetryDelayFunc: func(n int, err error, t *Task) time.Duration {
 				return time.Second
 			},
+			LogLevel: testLogLevel,
 		})
 		// Create a bunch of tasks
 		for i := 0; i < count; i++ {
@@ -90,8 +90,16 @@ func BenchmarkEndToEnd(b *testing.B) {
 		var wg sync.WaitGroup
 		wg.Add(count * 2)
 		handler := func(ctx context.Context, t *Task) error {
-			// randomly fail 1% of tasks
-			if rand.Intn(100) == 1 {
+			n, err := t.Payload.GetInt("data")
+			if err != nil {
+				b.Logf("internal error: %v", err)
+			}
+			retried, ok := GetRetryCount(ctx)
+			if !ok {
+				b.Logf("internal error: %v", err)
+			}
+			// Fail 1% of tasks for the first attempt.
+			if retried == 0 && n%100 == 0 {
 				return fmt.Errorf(":(")
 			}
 			wg.Done()
@@ -131,6 +139,7 @@ func BenchmarkEndToEndMultipleQueues(b *testing.B) {
 				"default": 3,
 				"low":     1,
 			},
+			LogLevel: testLogLevel,
 		})
 		// Create a bunch of tasks
 		for i := 0; i < highCount; i++ {
@@ -168,3 +177,62 @@ func BenchmarkEndToEndMultipleQueues(b *testing.B) {
 		b.StartTimer() // end teardown
 	}
 }
+
+// E2E benchmark to check client enqueue operation performs correctly,
+// while server is busy processing tasks.
+func BenchmarkClientWhileServerRunning(b *testing.B) {
+	const count = 10000
+	for n := 0; n < b.N; n++ {
+		b.StopTimer() // begin setup
+		setup(b)
+		redis := &RedisClientOpt{
+			Addr: redisAddr,
+			DB:   redisDB,
+		}
+		client := NewClient(redis)
+		srv := NewServer(redis, Config{
+			Concurrency: 10,
+			RetryDelayFunc: func(n int, err error, t *Task) time.Duration {
+				return time.Second
+			},
+			LogLevel: testLogLevel,
+		})
+		// Enqueue 10,000 tasks.
+		for i := 0; i < count; i++ {
+			t := NewTask(fmt.Sprintf("task%d", i), map[string]interface{}{"data": i})
+			if err := client.Enqueue(t); err != nil {
+				b.Fatalf("could not enqueue a task: %v", err)
+			}
+		}
+		// Schedule 10,000 tasks.
+		for i := 0; i < count; i++ {
+			t := NewTask(fmt.Sprintf("scheduled%d", i), map[string]interface{}{"data": i})
+			if err := client.EnqueueAt(time.Now().Add(time.Second), t); err != nil {
+				b.Fatalf("could not enqueue a task: %v", err)
+			}
+		}
+
+		handler := func(ctx context.Context, t *Task) error {
+			return nil
+		}
+		srv.Start(HandlerFunc(handler))
+
+		b.StartTimer() // end setup
+
+		b.Log("Starting enqueueing")
+		enqueued := 0
+		for enqueued < 100000 {
+			t := NewTask(fmt.Sprintf("enqueued%d", enqueued), map[string]interface{}{"data": enqueued})
+			if err := client.Enqueue(t); err != nil {
+				b.Logf("could not enqueue task %d: %v", enqueued, err)
+				continue
+			}
+			enqueued++
+		}
+		b.Logf("Finished enqueueing %d tasks", enqueued)
+
+		b.StopTimer() // begin teardown
+		srv.Stop()
+		b.StartTimer() // end teardown
+	}
+}

heartbeat.go

@@ -5,11 +5,13 @@
 package asynq
 
 import (
+	"os"
 	"sync"
 	"time"
 
 	"github.com/hibiken/asynq/internal/base"
 	"github.com/hibiken/asynq/internal/log"
+	"github.com/rs/xid"
 )
 
 // heartbeater is responsible for writing process info to redis periodically to
@@ -18,29 +20,69 @@ type heartbeater struct {
 	logger *log.Logger
 	broker base.Broker
 
-	ss *base.ServerState
-
 	// channel to communicate back to the long running "heartbeater" goroutine.
 	done chan struct{}
 
 	// interval between heartbeats.
 	interval time.Duration
+
+	// following fields are initialized at construction time and are immutable.
+	host           string
+	pid            int
+	serverID       string
+	concurrency    int
+	queues         map[string]int
+	strictPriority bool
+
+	// following fields are mutable and should be accessed only by the
+	// heartbeater goroutine. In other words, confine these variables
+	// to this goroutine only.
+	started time.Time
+	workers map[string]workerStat
+
+	// status is shared with other goroutine but is concurrency safe.
+	status *base.ServerStatus
+
+	// channels to receive updates on active workers.
+	starting <-chan *base.TaskMessage
+	finished <-chan *base.TaskMessage
 }
 
 type heartbeaterParams struct {
 	logger         *log.Logger
 	broker         base.Broker
-	serverState *base.ServerState
 	interval       time.Duration
+	concurrency    int
+	queues         map[string]int
+	strictPriority bool
+	status         *base.ServerStatus
+	starting       <-chan *base.TaskMessage
+	finished       <-chan *base.TaskMessage
 }
 
 func newHeartbeater(params heartbeaterParams) *heartbeater {
+	host, err := os.Hostname()
+	if err != nil {
+		host = "unknown-host"
+	}
+
 	return &heartbeater{
 		logger:   params.logger,
 		broker:   params.broker,
-		ss:       params.serverState,
 		done:     make(chan struct{}),
 		interval: params.interval,
+
+		host:           host,
+		pid:            os.Getpid(),
+		serverID:       xid.New().String(),
+		concurrency:    params.concurrency,
+		queues:         params.queues,
+		strictPriority: params.strictPriority,
+
+		status:   params.status,
+		workers:  make(map[string]workerStat),
+		starting: params.starting,
+		finished: params.finished,
 	}
 }
 
@@ -50,31 +92,74 @@ func (h *heartbeater) terminate() {
 	h.done <- struct{}{}
 }
 
+// A workerStat records the message a worker is working on
+// and the time the worker has started processing the message.
+type workerStat struct {
+	started time.Time
+	msg     *base.TaskMessage
+}
+
 func (h *heartbeater) start(wg *sync.WaitGroup) {
-	h.ss.SetStarted(time.Now())
-	h.ss.SetStatus(base.StatusRunning)
 	wg.Add(1)
 	go func() {
 		defer wg.Done()
+
+		h.started = time.Now()
+
 		h.beat()
+
+		timer := time.NewTimer(h.interval)
 		for {
 			select {
 			case <-h.done:
-				h.broker.ClearServerState(h.ss)
+				h.broker.ClearServerState(h.host, h.pid, h.serverID)
 				h.logger.Debug("Heartbeater done")
+				timer.Stop()
 				return
-			case <-time.After(h.interval):
+
+			case <-timer.C:
 				h.beat()
+				timer.Reset(h.interval)
+
+			case msg := <-h.starting:
+				h.workers[msg.ID.String()] = workerStat{time.Now(), msg}
+
+			case msg := <-h.finished:
+				delete(h.workers, msg.ID.String())
 			}
 		}
 	}()
 }
 
 func (h *heartbeater) beat() {
+	info := base.ServerInfo{
+		Host:              h.host,
+		PID:               h.pid,
+		ServerID:          h.serverID,
+		Concurrency:       h.concurrency,
+		Queues:            h.queues,
+		StrictPriority:    h.strictPriority,
+		Status:            h.status.String(),
+		Started:           h.started,
+		ActiveWorkerCount: len(h.workers),
+	}
+
+	var ws []*base.WorkerInfo
+	for id, stat := range h.workers {
+		ws = append(ws, &base.WorkerInfo{
+			Host:    h.host,
+			PID:     h.pid,
+			ID:      id,
+			Type:    stat.msg.Type,
+			Queue:   stat.msg.Queue,
+			Payload: stat.msg.Payload,
+			Started: stat.started,
+		})
+	}
+
 	// Note: Set TTL to be long enough so that it won't expire before we write again
 	// and short enough to expire quickly once the process is shut down or killed.
-	err := h.broker.WriteServerState(h.ss, h.interval*2)
-	if err != nil {
-		h.logger.Errorf("could not write heartbeat data: %v", err)
+	if err := h.broker.WriteServerState(&info, ws, h.interval*2); err != nil {
+		h.logger.Errorf("could not write server state data: %v", err)
 	}
 }

heartbeat_test.go

@@ -37,14 +37,24 @@ func TestHeartbeater(t *testing.T) {
 	for _, tc := range tests {
 		h.FlushDB(t, r)
 
-		state := base.NewServerState(tc.host, tc.pid, tc.concurrency, tc.queues, false)
+		status := base.NewServerStatus(base.StatusIdle)
 		hb := newHeartbeater(heartbeaterParams{
 			logger:         testLogger,
 			broker:         rdbClient,
-			serverState: state,
 			interval:       tc.interval,
+			concurrency:    tc.concurrency,
+			queues:         tc.queues,
+			strictPriority: false,
+			status:         status,
+			starting:       make(chan *base.TaskMessage),
+			finished:       make(chan *base.TaskMessage),
 		})
 
+		// Change host and pid fields for testing purpose.
+		hb.host = tc.host
+		hb.pid = tc.pid
+
+		status.Set(base.StatusRunning)
 		var wg sync.WaitGroup
 		hb.start(&wg)
 
@@ -80,7 +90,7 @@ func TestHeartbeater(t *testing.T) {
 		}
 
 		// status change
-		state.SetStatus(base.StatusStopped)
+		status.Set(base.StatusStopped)
 
 		// allow for heartbeater to write to redis
 		time.Sleep(tc.interval * 2)
@@ -119,12 +129,16 @@ func TestHeartbeaterWithRedisDown(t *testing.T) {
 	}()
 	r := rdb.NewRDB(setup(t))
 	testBroker := testbroker.NewTestBroker(r)
-	ss := base.NewServerState("localhost", 1234, 10, map[string]int{"default": 1}, false)
 	hb := newHeartbeater(heartbeaterParams{
 		logger:         testLogger,
 		broker:         testBroker,
-		serverState: ss,
 		interval:       time.Second,
+		concurrency:    10,
+		queues:         map[string]int{"default": 1},
+		strictPriority: false,
+		status:         base.NewServerStatus(base.StatusRunning),
+		starting:       make(chan *base.TaskMessage),
+		finished:       make(chan *base.TaskMessage),
 	})
 
 	testBroker.Sleep()

internal/asynqtest/asynqtest.go

@@ -57,7 +57,7 @@ var SortServerInfoOpt = cmp.Transformer("SortServerInfo", func(in []*base.Server
 var SortWorkerInfoOpt = cmp.Transformer("SortWorkerInfo", func(in []*base.WorkerInfo) []*base.WorkerInfo {
 	out := append([]*base.WorkerInfo(nil), in...) // Copy input to avoid mutating it
 	sort.Slice(out, func(i, j int) bool {
-		return out[i].ID.String() < out[j].ID.String()
+		return out[i].ID < out[j].ID
 	})
 	return out
 })

internal/base/base.go

@@ -7,6 +7,7 @@ package base
 
 import (
 	"context"
+	"encoding/json"
 	"fmt"
 	"strings"
 	"sync"
@@ -34,6 +35,7 @@ const (
 	RetryQueue      = "asynq:retry"                  // ZSET
 	DeadQueue       = "asynq:dead"                   // ZSET
 	InProgressQueue = "asynq:in_progress"            // LIST
+	PausedQueues    = "asynq:paused"                 // SET
 	CancelChannel   = "asynq:cancel"                 // PubSub channel
 )
 
@@ -105,28 +107,43 @@ type TaskMessage struct {
 	UniqueKey string
 }
 
-// ServerState holds process level information.
-//
-// ServerStates are safe for concurrent use by multiple goroutines.
-type ServerState struct {
-	mu             sync.Mutex // guards all data fields
-	id             xid.ID
-	concurrency    int
-	queues         map[string]int
-	strictPriority bool
-	pid            int
-	host           string
-	status         ServerStatus
-	started        time.Time
-	workers        map[string]*workerStats
+// EncodeMessage marshals the given task message in JSON and returns an encoded string.
+func EncodeMessage(msg *TaskMessage) (string, error) {
+	b, err := json.Marshal(msg)
+	if err != nil {
+		return "", err
+	}
+	return string(b), nil
+}
+
+// DecodeMessage unmarshals the given encoded string and returns a decoded task message.
+func DecodeMessage(s string) (*TaskMessage, error) {
+	d := json.NewDecoder(strings.NewReader(s))
+	d.UseNumber()
+	var msg TaskMessage
+	if err := d.Decode(&msg); err != nil {
+		return nil, err
+	}
+	return &msg, nil
 }
 
 // ServerStatus represents status of a server.
-type ServerStatus int
+// ServerStatus methods are concurrency safe.
+type ServerStatus struct {
+	mu  sync.Mutex
+	val ServerStatusValue
+}
+
+// NewServerStatus returns a new status instance given an initial value.
+func NewServerStatus(v ServerStatusValue) *ServerStatus {
+	return &ServerStatus{val: v}
+}
+
+type ServerStatusValue int
 
 const (
 	// StatusIdle indicates the server is in idle state.
-	StatusIdle ServerStatus = iota
+	StatusIdle ServerStatusValue = iota
 
 	// StatusRunning indicates the servier is up and processing tasks.
 	StatusRunning
@@ -145,117 +162,28 @@ var statuses = []string{
 	"stopped",
 }
 
-func (s ServerStatus) String() string {
-	if StatusIdle <= s && s <= StatusStopped {
-		return statuses[s]
+func (s *ServerStatus) String() string {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	if StatusIdle <= s.val && s.val <= StatusStopped {
+		return statuses[s.val]
 	}
 	return "unknown status"
 }
 
-type workerStats struct {
-	msg     *TaskMessage
-	started time.Time
+// Get returns the status value.
+func (s *ServerStatus) Get() ServerStatusValue {
+	s.mu.Lock()
+	v := s.val
+	s.mu.Unlock()
+	return v
 }
 
-// NewServerState returns a new instance of ServerState.
-func NewServerState(host string, pid, concurrency int, queues map[string]int, strict bool) *ServerState {
-	return &ServerState{
-		host:           host,
-		pid:            pid,
-		id:             xid.New(),
-		concurrency:    concurrency,
-		queues:         cloneQueueConfig(queues),
-		strictPriority: strict,
-		status:         StatusIdle,
-		workers:        make(map[string]*workerStats),
-	}
-}
-
-// SetStatus updates the status of server.
-func (ss *ServerState) SetStatus(status ServerStatus) {
-	ss.mu.Lock()
-	defer ss.mu.Unlock()
-	ss.status = status
-}
-
-// Status returns the status of server.
-func (ss *ServerState) Status() ServerStatus {
-	ss.mu.Lock()
-	defer ss.mu.Unlock()
-	return ss.status
-}
-
-// SetStarted records when the process started processing.
-func (ss *ServerState) SetStarted(t time.Time) {
-	ss.mu.Lock()
-	defer ss.mu.Unlock()
-	ss.started = t
-}
-
-// AddWorkerStats records when a worker started and which task it's processing.
-func (ss *ServerState) AddWorkerStats(msg *TaskMessage, started time.Time) {
-	ss.mu.Lock()
-	defer ss.mu.Unlock()
-	ss.workers[msg.ID.String()] = &workerStats{msg, started}
-}
-
-// DeleteWorkerStats removes a worker's entry from the process state.
-func (ss *ServerState) DeleteWorkerStats(msg *TaskMessage) {
-	ss.mu.Lock()
-	defer ss.mu.Unlock()
-	delete(ss.workers, msg.ID.String())
-}
-
-// GetInfo returns current state of server as a ServerInfo.
-func (ss *ServerState) GetInfo() *ServerInfo {
-	ss.mu.Lock()
-	defer ss.mu.Unlock()
-	return &ServerInfo{
-		Host:              ss.host,
-		PID:               ss.pid,
-		ServerID:          ss.id.String(),
-		Concurrency:       ss.concurrency,
-		Queues:            cloneQueueConfig(ss.queues),
-		StrictPriority:    ss.strictPriority,
-		Status:            ss.status.String(),
-		Started:           ss.started,
-		ActiveWorkerCount: len(ss.workers),
-	}
-}
-
-// GetWorkers returns a list of currently running workers' info.
-func (ss *ServerState) GetWorkers() []*WorkerInfo {
-	ss.mu.Lock()
-	defer ss.mu.Unlock()
-	var res []*WorkerInfo
-	for _, w := range ss.workers {
-		res = append(res, &WorkerInfo{
-			Host:    ss.host,
-			PID:     ss.pid,
-			ID:      w.msg.ID,
-			Type:    w.msg.Type,
-			Queue:   w.msg.Queue,
-			Payload: clonePayload(w.msg.Payload),
-			Started: w.started,
-		})
-	}
-	return res
-}
-
-func cloneQueueConfig(qcfg map[string]int) map[string]int {
-	res := make(map[string]int)
-	for qname, n := range qcfg {
-		res[qname] = n
-	}
-	return res
-}
-
-func clonePayload(payload map[string]interface{}) map[string]interface{} {
-	res := make(map[string]interface{})
-	for k, v := range payload {
-		res[k] = v
-	}
-	return res
+// Set sets the status value.
+func (s *ServerStatus) Set(v ServerStatusValue) {
+	s.mu.Lock()
+	s.val = v
+	s.mu.Unlock()
 }
 
 // ServerInfo holds information about a running server.
@@ -275,7 +203,7 @@ type ServerInfo struct {
 type WorkerInfo struct {
 	Host    string
 	PID     int
-	ID      xid.ID
+	ID      string
 	Type    string
 	Queue   string
 	Payload map[string]interface{}
@@ -343,10 +271,9 @@ type Broker interface {
 	ScheduleUnique(msg *TaskMessage, processAt time.Time, ttl time.Duration) error
 	Retry(msg *TaskMessage, processAt time.Time, errMsg string) error
 	Kill(msg *TaskMessage, errMsg string) error
-	RequeueAll() (int64, error)
-	CheckAndEnqueue(qnames ...string) error
-	WriteServerState(ss *ServerState, ttl time.Duration) error
-	ClearServerState(ss *ServerState) error
+	CheckAndEnqueue() error
+	WriteServerState(info *ServerInfo, workers []*WorkerInfo, ttl time.Duration) error
+	ClearServerState(host string, pid int, serverID string) error
 	CancelationPubSub() (*redis.PubSub, error) // TODO: Need to decouple from redis to support other brokers
 	PublishCancelation(id string) error
 	Close() error

internal/base/base_test.go

@@ -6,13 +6,12 @@ package base
 
 import (
 	"context"
-	"math/rand"
+	"encoding/json"
 	"sync"
 	"testing"
 	"time"
 
 	"github.com/google/go-cmp/cmp"
-	"github.com/google/go-cmp/cmp/cmpopts"
 	"github.com/rs/xid"
 )
 
@@ -108,69 +107,74 @@ func TestWorkersKey(t *testing.T) {
 	}
 }
 
-// Test for server state being accessed by multiple goroutines.
-// Run with -race flag to check for data race.
-func TestServerStateConcurrentAccess(t *testing.T) {
-	ss := NewServerState("127.0.0.1", 1234, 10, map[string]int{"default": 1}, false)
-	var wg sync.WaitGroup
-	started := time.Now()
-	msgs := []*TaskMessage{
-		{ID: xid.New(), Type: "type1", Payload: map[string]interface{}{"user_id": 42}},
-		{ID: xid.New(), Type: "type2"},
-		{ID: xid.New(), Type: "type3"},
+func TestMessageEncoding(t *testing.T) {
+	id := xid.New()
+	tests := []struct {
+		in  *TaskMessage
+		out *TaskMessage
+	}{
+		{
+			in: &TaskMessage{
+				Type:    "task1",
+				Payload: map[string]interface{}{"a": 1, "b": "hello!", "c": true},
+				ID:      id,
+				Queue:   "default",
+				Retry:   10,
+				Retried: 0,
+				Timeout: "0",
+			},
+			out: &TaskMessage{
+				Type:    "task1",
+				Payload: map[string]interface{}{"a": json.Number("1"), "b": "hello!", "c": true},
+				ID:      id,
+				Queue:   "default",
+				Retry:   10,
+				Retried: 0,
+				Timeout: "0",
+			},
+		},
 	}
 
-	// Simulate hearbeater calling SetStatus and SetStarted.
+	for _, tc := range tests {
+		encoded, err := EncodeMessage(tc.in)
+		if err != nil {
+			t.Errorf("EncodeMessage(msg) returned error: %v", err)
+			continue
+		}
+		decoded, err := DecodeMessage(encoded)
+		if err != nil {
+			t.Errorf("DecodeMessage(encoded) returned error: %v", err)
+			continue
+		}
+		if diff := cmp.Diff(tc.out, decoded); diff != "" {
+			t.Errorf("Decoded message == %+v, want %+v;(-want,+got)\n%s",
+				decoded, tc.out, diff)
+		}
+	}
+}
+
+// Test for status being accessed by multiple goroutines.
+// Run with -race flag to check for data race.
+func TestStatusConcurrentAccess(t *testing.T) {
+	status := NewServerStatus(StatusIdle)
+
+	var wg sync.WaitGroup
+
 	wg.Add(1)
 	go func() {
 		defer wg.Done()
-		ss.SetStarted(started)
-		ss.SetStatus(StatusRunning)
-		if status := ss.Status(); status != StatusRunning {
-			t.Errorf("(*ServerState).Status() = %v, want %v", status, StatusRunning)
-		}
+		status.Get()
+		status.String()
 	}()
 
-	// Simulate processor starting worker goroutines.
-	for _, msg := range msgs {
-		wg.Add(1)
-		ss.AddWorkerStats(msg, time.Now())
-		go func(msg *TaskMessage) {
-			defer wg.Done()
-			time.Sleep(time.Duration(rand.Intn(500)) * time.Millisecond)
-			ss.DeleteWorkerStats(msg)
-		}(msg)
-	}
-
-	// Simulate hearbeater calling Get and GetWorkers
 	wg.Add(1)
 	go func() {
-		wg.Done()
-		for i := 0; i < 5; i++ {
-			ss.GetInfo()
-			ss.GetWorkers()
-			time.Sleep(time.Duration(rand.Intn(100)) * time.Millisecond)
-		}
+		defer wg.Done()
+		status.Set(StatusStopped)
+		status.String()
 	}()
 
 	wg.Wait()
-
-	want := &ServerInfo{
-		Host:              "127.0.0.1",
-		PID:               1234,
-		Concurrency:       10,
-		Queues:            map[string]int{"default": 1},
-		StrictPriority:    false,
-		Status:            "running",
-		Started:           started,
-		ActiveWorkerCount: 0,
-	}
-
-	got := ss.GetInfo()
-	if diff := cmp.Diff(want, got, cmpopts.IgnoreFields(ServerInfo{}, "ServerID")); diff != "" {
-		t.Errorf("(*ServerState).GetInfo() = %+v, want %+v; (-want,+got)\n%s",
-			got, want, diff)
-	}
 }
 
 // Test for cancelations being accessed by multiple goroutines.

internal/rdb/inspect.go

@@ -7,6 +7,7 @@ package rdb
 import (
 	"encoding/json"
 	"fmt"
+	"sort"
 	"strings"
 	"time"
 
@@ -25,10 +26,24 @@ type Stats struct {
 	Dead       int
 	Processed  int
 	Failed     int
-	Queues     map[string]int // map of queue name to number of tasks in the queue (e.g., "default": 100, "critical": 20)
+	Queues     []*Queue
 	Timestamp  time.Time
 }
 
+// Queue represents a task queue.
+type Queue struct {
+	// Name of the queue (e.g. "default", "critical").
+	// Note: It doesn't include the prefix "asynq:queues:".
+	Name string
+
+	// Paused indicates whether the queue is paused.
+	// If true, tasks in the queue should not be processed.
+	Paused bool
+
+	// Size is the number of tasks in the queue.
+	Size int
+}
+
 // DailyStats holds aggregate data for a given day.
 type DailyStats struct {
 	Processed int
@@ -143,8 +158,12 @@ func (r *RDB) CurrentStats() (*Stats, error) {
 	if err != nil {
 		return nil, err
 	}
+	paused, err := r.client.SMembersMap(base.PausedQueues).Result()
+	if err != nil {
+		return nil, err
+	}
 	stats := &Stats{
-		Queues:    make(map[string]int),
+		Queues:    make([]*Queue, 0),
 		Timestamp: now,
 	}
 	for i := 0; i < len(data); i += 2 {
@@ -154,7 +173,14 @@ func (r *RDB) CurrentStats() (*Stats, error) {
 		switch {
 		case strings.HasPrefix(key, base.QueuePrefix):
 			stats.Enqueued += val
-			stats.Queues[strings.TrimPrefix(key, base.QueuePrefix)] = val
+			q := Queue{
+				Name: strings.TrimPrefix(key, base.QueuePrefix),
+				Size: val,
+			}
+			if _, exist := paused[key]; exist {
+				q.Paused = true
+			}
+			stats.Queues = append(stats.Queues, &q)
 		case key == base.InProgressQueue:
 			stats.InProgress = val
 		case key == base.ScheduledQueue:
@@ -169,6 +195,9 @@ func (r *RDB) CurrentStats() (*Stats, error) {
 			stats.Failed = val
 		}
 	}
+	sort.Slice(stats.Queues, func(i, j int) bool {
+		return stats.Queues[i].Name < stats.Queues[j].Name
+	})
 	return stats, nil
 }
 
@@ -830,3 +859,33 @@ func (r *RDB) ListWorkers() ([]*base.WorkerInfo, error) {
 	}
 	return workers, nil
 }
+
+// KEYS[1] -> asynq:paused
+// ARGV[1] -> asynq:queues:<qname> - queue to pause
+var pauseCmd = redis.NewScript(`
+local ismem = redis.call("SISMEMBER", KEYS[1], ARGV[1])
+if ismem == 1 then
+	return redis.error_reply("queue is already paused")
+end
+return redis.call("SADD", KEYS[1], ARGV[1])`)
+
+// Pause pauses processing of tasks from the given queue.
+func (r *RDB) Pause(qname string) error {
+	qkey := base.QueueKey(qname)
+	return pauseCmd.Run(r.client, []string{base.PausedQueues}, qkey).Err()
+}
+
+// KEYS[1] -> asynq:paused
+// ARGV[1] -> asynq:queues:<qname> - queue to unpause
+var unpauseCmd = redis.NewScript(`
+local ismem = redis.call("SISMEMBER", KEYS[1], ARGV[1])
+if ismem == 0 then
+	return redis.error_reply("queue is not paused")
+end
+return redis.call("SREM", KEYS[1], ARGV[1])`)
+
+// Unpause resumes processing of tasks from the given queue.
+func (r *RDB) Unpause(qname string) error {
+	qkey := base.QueueKey(qname)
+	return unpauseCmd.Run(r.client, []string{base.PausedQueues}, qkey).Err()
+}

internal/rdb/inspect_test.go

@@ -38,6 +38,7 @@ func TestCurrentStats(t *testing.T) {
 		processed  int
 		failed     int
 		allQueues  []interface{}
+		paused     []string
 		want       *Stats
 	}{
 		{
@@ -55,6 +56,7 @@ func TestCurrentStats(t *testing.T) {
 			processed: 120,
 			failed:    2,
 			allQueues: []interface{}{base.DefaultQueue, base.QueueKey("critical"), base.QueueKey("low")},
+			paused:    []string{},
 			want: &Stats{
 				Enqueued:   3,
 				InProgress: 1,
@@ -64,7 +66,12 @@ func TestCurrentStats(t *testing.T) {
 				Processed:  120,
 				Failed:     2,
 				Timestamp:  now,
-				Queues:     map[string]int{base.DefaultQueueName: 1, "critical": 1, "low": 1},
+				// Queues should be sorted by name.
+				Queues: []*Queue{
+					{Name: "critical", Paused: false, Size: 1},
+					{Name: "default", Paused: false, Size: 1},
+					{Name: "low", Paused: false, Size: 1},
+				},
 			},
 		},
 		{
@@ -82,6 +89,7 @@ func TestCurrentStats(t *testing.T) {
 			processed: 90,
 			failed:    10,
 			allQueues: []interface{}{base.DefaultQueue},
+			paused:    []string{},
 			want: &Stats{
 				Enqueued:   0,
 				InProgress: 0,
@@ -91,13 +99,52 @@ func TestCurrentStats(t *testing.T) {
 				Processed:  90,
 				Failed:     10,
 				Timestamp:  now,
-				Queues:     map[string]int{base.DefaultQueueName: 0},
+				Queues: []*Queue{
+					{Name: base.DefaultQueueName, Paused: false, Size: 0},
+				},
+			},
+		},
+		{
+			enqueued: map[string][]*base.TaskMessage{
+				base.DefaultQueueName: {m1},
+				"critical":            {m5},
+				"low":                 {m6},
+			},
+			inProgress: []*base.TaskMessage{m2},
+			scheduled: []h.ZSetEntry{
+				{Msg: m3, Score: float64(now.Add(time.Hour).Unix())},
+				{Msg: m4, Score: float64(now.Unix())}},
+			retry:     []h.ZSetEntry{},
+			dead:      []h.ZSetEntry{},
+			processed: 120,
+			failed:    2,
+			allQueues: []interface{}{base.DefaultQueue, base.QueueKey("critical"), base.QueueKey("low")},
+			paused:    []string{"critical", "low"},
+			want: &Stats{
+				Enqueued:   3,
+				InProgress: 1,
+				Scheduled:  2,
+				Retry:      0,
+				Dead:       0,
+				Processed:  120,
+				Failed:     2,
+				Timestamp:  now,
+				Queues: []*Queue{
+					{Name: "critical", Paused: true, Size: 1},
+					{Name: "default", Paused: false, Size: 1},
+					{Name: "low", Paused: true, Size: 1},
+				},
 			},
 		},
 	}
 
 	for _, tc := range tests {
 		h.FlushDB(t, r.client) // clean up db before each test case
+		for _, qname := range tc.paused {
+			if err := r.Pause(qname); err != nil {
+				t.Fatal(err)
+			}
+		}
 		for qname, msgs := range tc.enqueued {
 			h.SeedEnqueuedQueue(t, r.client, msgs, qname)
 		}
@@ -136,7 +183,7 @@ func TestCurrentStatsWithoutData(t *testing.T) {
 		Processed:  0,
 		Failed:     0,
 		Timestamp:  time.Now(),
-		Queues:     map[string]int{},
+		Queues:     make([]*Queue, 0),
 	}
 
 	got, err := r.CurrentStats()
@@ -658,12 +705,14 @@ func TestListRetry(t *testing.T) {
 func TestListRetryPagination(t *testing.T) {
 	r := setup(t)
 	// create 100 tasks with an increasing number of wait time.
+	now := time.Now()
+	var seed []h.ZSetEntry
 	for i := 0; i < 100; i++ {
 		msg := h.NewTaskMessage(fmt.Sprintf("task %d", i), nil)
-		if err := r.Retry(msg, time.Now().Add(time.Duration(i)*time.Second), "error"); err != nil {
-			t.Fatal(err)
-		}
+		processAt := now.Add(time.Duration(i) * time.Second)
+		seed = append(seed, h.ZSetEntry{Msg: msg, Score: float64(processAt.Unix())})
 	}
+	h.SeedRetryQueue(t, r.client, seed)
 
 	tests := []struct {
 		desc      string
@@ -2055,60 +2104,48 @@ func TestListServers(t *testing.T) {
 	r := setup(t)
 
 	started1 := time.Now().Add(-time.Hour)
-	ss1 := base.NewServerState("do.droplet1", 1234, 10, map[string]int{"default": 1}, false)
-	ss1.SetStarted(started1)
-	ss1.SetStatus(base.StatusRunning)
 	info1 := &base.ServerInfo{
-		Concurrency:       10,
-		Queues:            map[string]int{"default": 1},
 		Host:              "do.droplet1",
 		PID:               1234,
+		ServerID:          "server123",
+		Concurrency:       10,
+		Queues:            map[string]int{"default": 1},
 		Status:            "running",
 		Started:           started1,
 		ActiveWorkerCount: 0,
 	}
 
 	started2 := time.Now().Add(-2 * time.Hour)
-	ss2 := base.NewServerState("do.droplet2", 9876, 20, map[string]int{"email": 1}, false)
-	ss2.SetStarted(started2)
-	ss2.SetStatus(base.StatusStopped)
-	ss2.AddWorkerStats(h.NewTaskMessage("send_email", nil), time.Now())
 	info2 := &base.ServerInfo{
-		Concurrency:       20,
-		Queues:            map[string]int{"email": 1},
 		Host:              "do.droplet2",
 		PID:               9876,
+		ServerID:          "server456",
+		Concurrency:       20,
+		Queues:            map[string]int{"email": 1},
 		Status:            "stopped",
 		Started:           started2,
 		ActiveWorkerCount: 1,
 	}
 
 	tests := []struct {
-		serverStates []*base.ServerState
-		want         []*base.ServerInfo
+		data []*base.ServerInfo
 	}{
 		{
-			serverStates: []*base.ServerState{},
-			want:         []*base.ServerInfo{},
+			data: []*base.ServerInfo{},
 		},
 		{
-			serverStates: []*base.ServerState{ss1},
-			want:         []*base.ServerInfo{info1},
+			data: []*base.ServerInfo{info1},
 		},
 		{
-			serverStates: []*base.ServerState{ss1, ss2},
-			want:         []*base.ServerInfo{info1, info2},
+			data: []*base.ServerInfo{info1, info2},
 		},
 	}
 
-	ignoreOpt := cmpopts.IgnoreUnexported(base.ServerInfo{})
-	ignoreFieldOpt := cmpopts.IgnoreFields(base.ServerInfo{}, "ServerID")
-
 	for _, tc := range tests {
 		h.FlushDB(t, r.client)
 
-		for _, ss := range tc.serverStates {
-			if err := r.WriteServerState(ss, 5*time.Second); err != nil {
+		for _, info := range tc.data {
+			if err := r.WriteServerState(info, []*base.WorkerInfo{}, 5*time.Second); err != nil {
 				t.Fatal(err)
 			}
 		}
@@ -2117,9 +2154,9 @@ func TestListServers(t *testing.T) {
 		if err != nil {
 			t.Errorf("r.ListServers returned an error: %v", err)
 		}
-		if diff := cmp.Diff(tc.want, got, h.SortServerInfoOpt, ignoreOpt, ignoreFieldOpt); diff != "" {
+		if diff := cmp.Diff(tc.data, got, h.SortServerInfoOpt); diff != "" {
 			t.Errorf("r.ListServers returned %v, want %v; (-want,+got)\n%s",
-				got, tc.serverStates, diff)
+				got, tc.data, diff)
 		}
 	}
 }
@@ -2127,37 +2164,23 @@ func TestListServers(t *testing.T) {
 func TestListWorkers(t *testing.T) {
 	r := setup(t)
 
-	const (
+	var (
 		host = "127.0.0.1"
 		pid  = 4567
+
+		m1 = h.NewTaskMessage("send_email", map[string]interface{}{"user_id": "abc123"})
+		m2 = h.NewTaskMessage("gen_thumbnail", map[string]interface{}{"path": "some/path/to/image/file"})
+		m3 = h.NewTaskMessage("reindex", map[string]interface{}{})
 	)
 
-	m1 := h.NewTaskMessage("send_email", map[string]interface{}{"user_id": "abc123"})
-	m2 := h.NewTaskMessage("gen_thumbnail", map[string]interface{}{"path": "some/path/to/image/file"})
-	m3 := h.NewTaskMessage("reindex", map[string]interface{}{})
-	t1 := time.Now().Add(-time.Second)
-	t2 := time.Now().Add(-10 * time.Second)
-	t3 := time.Now().Add(-time.Minute)
-
-	type workerStats struct {
-		msg     *base.TaskMessage
-		started time.Time
-	}
-
 	tests := []struct {
-		workers []*workerStats
-		want    []*base.WorkerInfo
+		data []*base.WorkerInfo
 	}{
 		{
-			workers: []*workerStats{
-				{m1, t1},
-				{m2, t2},
-				{m3, t3},
-			},
-			want: []*base.WorkerInfo{
-				{Host: host, PID: pid, ID: m1.ID, Type: m1.Type, Queue: m1.Queue, Payload: m1.Payload, Started: t1},
-				{Host: host, PID: pid, ID: m2.ID, Type: m2.Type, Queue: m2.Queue, Payload: m2.Payload, Started: t2},
-				{Host: host, PID: pid, ID: m3.ID, Type: m3.Type, Queue: m3.Queue, Payload: m3.Payload, Started: t3},
+			data: []*base.WorkerInfo{
+				{Host: host, PID: pid, ID: m1.ID.String(), Type: m1.Type, Queue: m1.Queue, Payload: m1.Payload, Started: time.Now().Add(-1 * time.Second)},
+				{Host: host, PID: pid, ID: m2.ID.String(), Type: m2.Type, Queue: m2.Queue, Payload: m2.Payload, Started: time.Now().Add(-5 * time.Second)},
+				{Host: host, PID: pid, ID: m3.ID.String(), Type: m3.Type, Queue: m3.Queue, Payload: m3.Payload, Started: time.Now().Add(-30 * time.Second)},
 			},
 		},
 	}
@@ -2165,13 +2188,7 @@ func TestListWorkers(t *testing.T) {
 	for _, tc := range tests {
 		h.FlushDB(t, r.client)
 
-		ss := base.NewServerState(host, pid, 10, map[string]int{"default": 1}, false)
-
-		for _, w := range tc.workers {
-			ss.AddWorkerStats(w.msg, w.started)
-		}
-
-		err := r.WriteServerState(ss, time.Minute)
+		err := r.WriteServerState(&base.ServerInfo{}, tc.data, time.Minute)
 		if err != nil {
 			t.Errorf("could not write server state to redis: %v", err)
 			continue
@@ -2183,8 +2200,165 @@ func TestListWorkers(t *testing.T) {
 			continue
 		}
 
-		if diff := cmp.Diff(tc.want, got, h.SortWorkerInfoOpt); diff != "" {
-			t.Errorf("(*RDB).ListWorkers() = %v, want = %v; (-want,+got)\n%s", got, tc.want, diff)
+		if diff := cmp.Diff(tc.data, got, h.SortWorkerInfoOpt); diff != "" {
+			t.Errorf("(*RDB).ListWorkers() = %v, want = %v; (-want,+got)\n%s", got, tc.data, diff)
+		}
+	}
+}
+
+func TestPause(t *testing.T) {
+	r := setup(t)
+
+	tests := []struct {
+		initial []string // initial keys in the paused set
+		qname   string   // name of the queue to pause
+		want    []string // expected keys in the paused set
+	}{
+		{[]string{}, "default", []string{"asynq:queues:default"}},
+		{[]string{"asynq:queues:default"}, "critical", []string{"asynq:queues:default", "asynq:queues:critical"}},
+	}
+
+	for _, tc := range tests {
+		h.FlushDB(t, r.client)
+
+		// Set up initial state.
+		for _, qkey := range tc.initial {
+			if err := r.client.SAdd(base.PausedQueues, qkey).Err(); err != nil {
+				t.Fatal(err)
+			}
+		}
+
+		err := r.Pause(tc.qname)
+		if err != nil {
+			t.Errorf("Pause(%q) returned error: %v", tc.qname, err)
+		}
+
+		got, err := r.client.SMembers(base.PausedQueues).Result()
+		if err != nil {
+			t.Fatal(err)
+		}
+
+		if diff := cmp.Diff(tc.want, got, h.SortStringSliceOpt); diff != "" {
+			t.Errorf("%q has members %v, want %v; (-want,+got)\n%s",
+				base.PausedQueues, got, tc.want, diff)
+		}
+	}
+}
+
+func TestPauseError(t *testing.T) {
+	r := setup(t)
+
+	tests := []struct {
+		desc    string   // test case description
+		initial []string // initial keys in the paused set
+		qname   string   // name of the queue to pause
+		want    []string // expected keys in the paused set
+	}{
+		{"queue already paused", []string{"asynq:queues:default"}, "default", []string{"asynq:queues:default"}},
+	}
+
+	for _, tc := range tests {
+		h.FlushDB(t, r.client)
+
+		// Set up initial state.
+		for _, qkey := range tc.initial {
+			if err := r.client.SAdd(base.PausedQueues, qkey).Err(); err != nil {
+				t.Fatal(err)
+			}
+		}
+
+		err := r.Pause(tc.qname)
+		if err == nil {
+			t.Errorf("%s; Pause(%q) returned nil: want error", tc.desc, tc.qname)
+		}
+
+		got, err := r.client.SMembers(base.PausedQueues).Result()
+		if err != nil {
+			t.Fatal(err)
+		}
+
+		if diff := cmp.Diff(tc.want, got, h.SortStringSliceOpt); diff != "" {
+			t.Errorf("%s; %q has members %v, want %v; (-want,+got)\n%s",
+				tc.desc, base.PausedQueues, got, tc.want, diff)
+		}
+	}
+}
+
+func TestUnpause(t *testing.T) {
+	r := setup(t)
+
+	tests := []struct {
+		initial []string // initial keys in the paused set
+		qname   string   // name of the queue to unpause
+		want    []string // expected keys in the paused set
+	}{
+		{[]string{"asynq:queues:default"}, "default", []string{}},
+		{[]string{"asynq:queues:default", "asynq:queues:low"}, "low", []string{"asynq:queues:default"}},
+	}
+
+	for _, tc := range tests {
+		h.FlushDB(t, r.client)
+
+		// Set up initial state.
+		for _, qkey := range tc.initial {
+			if err := r.client.SAdd(base.PausedQueues, qkey).Err(); err != nil {
+				t.Fatal(err)
+			}
+		}
+
+		err := r.Unpause(tc.qname)
+		if err != nil {
+			t.Errorf("Unpause(%q) returned error: %v", tc.qname, err)
+		}
+
+		got, err := r.client.SMembers(base.PausedQueues).Result()
+		if err != nil {
+			t.Fatal(err)
+		}
+
+		if diff := cmp.Diff(tc.want, got, h.SortStringSliceOpt); diff != "" {
+			t.Errorf("%q has members %v, want %v; (-want,+got)\n%s",
+				base.PausedQueues, got, tc.want, diff)
+		}
+	}
+}
+
+func TestUnpauseError(t *testing.T) {
+	r := setup(t)
+
+	tests := []struct {
+		desc    string   // test case description
+		initial []string // initial keys in the paused set
+		qname   string   // name of the queue to unpause
+		want    []string // expected keys in the paused set
+	}{
+		{"set is empty", []string{}, "default", []string{}},
+		{"queue is not in the set", []string{"asynq:queues:default"}, "low", []string{"asynq:queues:default"}},
+	}
+
+	for _, tc := range tests {
+		h.FlushDB(t, r.client)
+
+		// Set up initial state.
+		for _, qkey := range tc.initial {
+			if err := r.client.SAdd(base.PausedQueues, qkey).Err(); err != nil {
+				t.Fatal(err)
+			}
+		}
+
+		err := r.Unpause(tc.qname)
+		if err == nil {
+			t.Errorf("%s; Unpause(%q) returned nil: want error", tc.desc, tc.qname)
+		}
+
+		got, err := r.client.SMembers(base.PausedQueues).Result()
+		if err != nil {
+			t.Fatal(err)
+		}
+
+		if diff := cmp.Diff(tc.want, got, h.SortStringSliceOpt); diff != "" {
+			t.Errorf("%s; %q has members %v, want %v; (-want,+got)\n%s",
+				tc.desc, base.PausedQueues, got, tc.want, diff)
 		}
 	}
 }

internal/rdb/rdb.go

@@ -54,12 +54,12 @@ return 1`)
 
 // Enqueue inserts the given task to the tail of the queue.
 func (r *RDB) Enqueue(msg *base.TaskMessage) error {
-	bytes, err := json.Marshal(msg)
+	encoded, err := base.EncodeMessage(msg)
 	if err != nil {
 		return err
 	}
 	key := base.QueueKey(msg.Queue)
-	return enqueueCmd.Run(r.client, []string{key, base.AllQueues}, bytes).Err()
+	return enqueueCmd.Run(r.client, []string{key, base.AllQueues}, encoded).Err()
 }
 
 // KEYS[1] -> unique key in the form <type>:<payload>:<qname>
@@ -81,14 +81,14 @@ return 1
 // EnqueueUnique inserts the given task if the task's uniqueness lock can be acquired.
 // It returns ErrDuplicateTask if the lock cannot be acquired.
 func (r *RDB) EnqueueUnique(msg *base.TaskMessage, ttl time.Duration) error {
-	bytes, err := json.Marshal(msg)
+	encoded, err := base.EncodeMessage(msg)
 	if err != nil {
 		return err
 	}
 	key := base.QueueKey(msg.Queue)
 	res, err := enqueueUniqueCmd.Run(r.client,
 		[]string{msg.UniqueKey, key, base.AllQueues},
-		msg.ID.String(), int(ttl.Seconds()), bytes).Result()
+		msg.ID.String(), int(ttl.Seconds()), encoded).Result()
 	if err != nil {
 		return err
 	}
@@ -103,56 +103,43 @@ func (r *RDB) EnqueueUnique(msg *base.TaskMessage, ttl time.Duration) error {
 }
 
 // Dequeue queries given queues in order and pops a task message if there is one and returns it.
+// Dequeue skips a queue if the queue is paused.
 // If all queues are empty, ErrNoProcessableTask error is returned.
 func (r *RDB) Dequeue(qnames ...string) (*base.TaskMessage, error) {
-	var data string
-	var err error
-	if len(qnames) == 1 {
-		data, err = r.dequeueSingle(base.QueueKey(qnames[0]))
-	} else {
-		var keys []string
+	var qkeys []interface{}
 	for _, q := range qnames {
-			keys = append(keys, base.QueueKey(q))
-		}
-		data, err = r.dequeue(keys...)
+		qkeys = append(qkeys, base.QueueKey(q))
 	}
+	data, err := r.dequeue(qkeys...)
 	if err == redis.Nil {
 		return nil, ErrNoProcessableTask
 	}
 	if err != nil {
 		return nil, err
 	}
-	var msg base.TaskMessage
-	err = json.Unmarshal([]byte(data), &msg)
-	if err != nil {
-		return nil, err
-	}
-	return &msg, nil
-}
-
-func (r *RDB) dequeueSingle(queue string) (data string, err error) {
-	// timeout needed to avoid blocking forever
-	return r.client.BRPopLPush(queue, base.InProgressQueue, time.Second).Result()
+	return base.DecodeMessage(data)
 }
 
 // KEYS[1] -> asynq:in_progress
+// KEYS[2] -> asynq:paused
 // ARGV    -> List of queues to query in order
+//
+// dequeueCmd checks whether a queue is paused first, before
+// calling RPOPLPUSH to pop a task from the queue.
 var dequeueCmd = redis.NewScript(`
-local res
 for _, qkey in ipairs(ARGV) do
-	res = redis.call("RPOPLPUSH", qkey, KEYS[1])
+	if redis.call("SISMEMBER", KEYS[2], qkey) == 0 then
+		local res = redis.call("RPOPLPUSH", qkey, KEYS[1])
 		if res then
 			return res
 		end
+	end
 end
-return res`)
+return nil`)
 
-func (r *RDB) dequeue(queues ...string) (data string, err error) {
-	var args []interface{}
-	for _, qkey := range queues {
-		args = append(args, qkey)
-	}
-	res, err := dequeueCmd.Run(r.client, []string{base.InProgressQueue}, args...).Result()
+func (r *RDB) dequeue(qkeys ...interface{}) (data string, err error) {
+	res, err := dequeueCmd.Run(r.client,
+		[]string{base.InProgressQueue, base.PausedQueues}, qkeys...).Result()
 	if err != nil {
 		return "", err
 	}
@@ -167,7 +154,10 @@ func (r *RDB) dequeue(queues ...string) (data string, err error) {
 // ARGV[3] -> task ID
 // Note: LREM count ZERO means "remove all elements equal to val"
 var doneCmd = redis.NewScript(`
-redis.call("LREM", KEYS[1], 0, ARGV[1]) 
+local x = redis.call("LREM", KEYS[1], 0, ARGV[1]) 
+if x == 0 then
+  return redis.error_reply("NOT FOUND")
+end
 local n = redis.call("INCR", KEYS[2])
 if tonumber(n) == 1 then
 	redis.call("EXPIREAT", KEYS[2], ARGV[2])
@@ -181,7 +171,7 @@ return redis.status_reply("OK")
 // Done removes the task from in-progress queue to mark the task as done.
 // It removes a uniqueness lock acquired by the task, if any.
 func (r *RDB) Done(msg *base.TaskMessage) error {
-	bytes, err := json.Marshal(msg)
+	encoded, err := base.EncodeMessage(msg)
 	if err != nil {
 		return err
 	}
@@ -190,7 +180,7 @@ func (r *RDB) Done(msg *base.TaskMessage) error {
 	expireAt := now.Add(statsTTL)
 	return doneCmd.Run(r.client,
 		[]string{base.InProgressQueue, processedKey, msg.UniqueKey},
-		bytes, expireAt.Unix(), msg.ID.String()).Err()
+		encoded, expireAt.Unix(), msg.ID.String()).Err()
 }
 
 // KEYS[1] -> asynq:in_progress
@@ -204,13 +194,13 @@ return redis.status_reply("OK")`)
 
 // Requeue moves the task from in-progress queue to the specified queue.
 func (r *RDB) Requeue(msg *base.TaskMessage) error {
-	bytes, err := json.Marshal(msg)
+	encoded, err := base.EncodeMessage(msg)
 	if err != nil {
 		return err
 	}
 	return requeueCmd.Run(r.client,
 		[]string{base.InProgressQueue, base.QueueKey(msg.Queue)},
-		string(bytes)).Err()
+		encoded).Err()
 }
 
 // KEYS[1] -> asynq:scheduled
@@ -226,7 +216,7 @@ return 1
 
 // Schedule adds the task to the backlog queue to be processed in the future.
 func (r *RDB) Schedule(msg *base.TaskMessage, processAt time.Time) error {
-	bytes, err := json.Marshal(msg)
+	encoded, err := base.EncodeMessage(msg)
 	if err != nil {
 		return err
 	}
@@ -234,7 +224,7 @@ func (r *RDB) Schedule(msg *base.TaskMessage, processAt time.Time) error {
 	score := float64(processAt.Unix())
 	return scheduleCmd.Run(r.client,
 		[]string{base.ScheduledQueue, base.AllQueues},
-		score, bytes, qkey).Err()
+		score, encoded, qkey).Err()
 }
 
 // KEYS[1] -> unique key in the format <type>:<payload>:<qname>
@@ -258,7 +248,7 @@ return 1
 // ScheduleUnique adds the task to the backlog queue to be processed in the future if the uniqueness lock can be acquired.
 // It returns ErrDuplicateTask if the lock cannot be acquired.
 func (r *RDB) ScheduleUnique(msg *base.TaskMessage, processAt time.Time, ttl time.Duration) error {
-	bytes, err := json.Marshal(msg)
+	encoded, err := base.EncodeMessage(msg)
 	if err != nil {
 		return err
 	}
@@ -266,7 +256,7 @@ func (r *RDB) ScheduleUnique(msg *base.TaskMessage, processAt time.Time, ttl tim
 	score := float64(processAt.Unix())
 	res, err := scheduleUniqueCmd.Run(r.client,
 		[]string{msg.UniqueKey, base.ScheduledQueue, base.AllQueues},
-		msg.ID.String(), int(ttl.Seconds()), score, bytes, qkey).Result()
+		msg.ID.String(), int(ttl.Seconds()), score, encoded, qkey).Result()
 	if err != nil {
 		return err
 	}
@@ -289,7 +279,10 @@ func (r *RDB) ScheduleUnique(msg *base.TaskMessage, processAt time.Time, ttl tim
 // ARGV[3] -> retry_at UNIX timestamp
 // ARGV[4] -> stats expiration timestamp
 var retryCmd = redis.NewScript(`
-redis.call("LREM", KEYS[1], 0, ARGV[1])
+local x = redis.call("LREM", KEYS[1], 0, ARGV[1])
+if x == 0 then
+  return redis.error_reply("NOT FOUND")
+end
 redis.call("ZADD", KEYS[2], ARGV[3], ARGV[2])
 local n = redis.call("INCR", KEYS[3])
 if tonumber(n) == 1 then
@@ -304,14 +297,14 @@ return redis.status_reply("OK")`)
 // Retry moves the task from in-progress to retry queue, incrementing retry count
 // and assigning error message to the task message.
 func (r *RDB) Retry(msg *base.TaskMessage, processAt time.Time, errMsg string) error {
-	bytesToRemove, err := json.Marshal(msg)
+	msgToRemove, err := base.EncodeMessage(msg)
 	if err != nil {
 		return err
 	}
 	modified := *msg
 	modified.Retried++
 	modified.ErrorMsg = errMsg
-	bytesToAdd, err := json.Marshal(&modified)
+	msgToAdd, err := base.EncodeMessage(&modified)
 	if err != nil {
 		return err
 	}
@@ -321,7 +314,7 @@ func (r *RDB) Retry(msg *base.TaskMessage, processAt time.Time, errMsg string) e
 	expireAt := now.Add(statsTTL)
 	return retryCmd.Run(r.client,
 		[]string{base.InProgressQueue, base.RetryQueue, processedKey, failureKey},
-		string(bytesToRemove), string(bytesToAdd), processAt.Unix(), expireAt.Unix()).Err()
+		msgToRemove, msgToAdd, processAt.Unix(), expireAt.Unix()).Err()
 }
 
 const (
@@ -340,7 +333,10 @@ const (
 // ARGV[5] -> max number of tasks in dead queue (e.g., 100)
 // ARGV[6] -> stats expiration timestamp
 var killCmd = redis.NewScript(`
-redis.call("LREM", KEYS[1], 0, ARGV[1])
+local x = redis.call("LREM", KEYS[1], 0, ARGV[1])
+if x == 0 then
+  return redis.error_reply("NOT FOUND")
+end
 redis.call("ZADD", KEYS[2], ARGV[3], ARGV[2])
 redis.call("ZREMRANGEBYSCORE", KEYS[2], "-inf", ARGV[4])
 redis.call("ZREMRANGEBYRANK", KEYS[2], 0, -ARGV[5])
@@ -358,13 +354,13 @@ return redis.status_reply("OK")`)
 // the error message to the task.
 // It also trims the set by timestamp and set size.
 func (r *RDB) Kill(msg *base.TaskMessage, errMsg string) error {
-	bytesToRemove, err := json.Marshal(msg)
+	msgToRemove, err := base.EncodeMessage(msg)
 	if err != nil {
 		return err
 	}
 	modified := *msg
 	modified.ErrorMsg = errMsg
-	bytesToAdd, err := json.Marshal(&modified)
+	msgToAdd, err := base.EncodeMessage(&modified)
 	if err != nil {
 		return err
 	}
@@ -375,105 +371,62 @@ func (r *RDB) Kill(msg *base.TaskMessage, errMsg string) error {
 	expireAt := now.Add(statsTTL)
 	return killCmd.Run(r.client,
 		[]string{base.InProgressQueue, base.DeadQueue, processedKey, failureKey},
-		string(bytesToRemove), string(bytesToAdd), now.Unix(), limit, maxDeadTasks, expireAt.Unix()).Err()
+		msgToRemove, msgToAdd, now.Unix(), limit, maxDeadTasks, expireAt.Unix()).Err()
 }
 
-// KEYS[1] -> asynq:in_progress
-// ARGV[1] -> queue prefix
-var requeueAllCmd = redis.NewScript(`
-local msgs = redis.call("LRANGE", KEYS[1], 0, -1)
-for _, msg in ipairs(msgs) do
-	local decoded = cjson.decode(msg)
-	local qkey = ARGV[1] .. decoded["Queue"]
-	redis.call("RPUSH", qkey, msg)
-	redis.call("LREM", KEYS[1], 0, msg)
-end
-return table.getn(msgs)`)
-
-// RequeueAll moves all tasks from in-progress list to the queue
-// and reports the number of tasks restored.
-func (r *RDB) RequeueAll() (int64, error) {
-	res, err := requeueAllCmd.Run(r.client, []string{base.InProgressQueue}, base.QueuePrefix).Result()
-	if err != nil {
-		return 0, err
-	}
-	n, ok := res.(int64)
-	if !ok {
-		return 0, fmt.Errorf("could not cast %v to int64", res)
-	}
-	return n, nil
-}
-
-// CheckAndEnqueue checks for all scheduled tasks and enqueues any tasks that
-// have to be processed.
-//
-// qnames specifies to which queues to send tasks.
-func (r *RDB) CheckAndEnqueue(qnames ...string) error {
+// CheckAndEnqueue checks for all scheduled/retry tasks and enqueues any tasks that
+// are ready to be processed.
+func (r *RDB) CheckAndEnqueue() (err error) {
 	delayed := []string{base.ScheduledQueue, base.RetryQueue}
 	for _, zset := range delayed {
-		var err error
-		if len(qnames) == 1 {
-			err = r.forwardSingle(zset, base.QueueKey(qnames[0]))
-		} else {
-			err = r.forward(zset)
-		}
+		n := 1
+		for n != 0 {
+			n, err = r.forward(zset)
 			if err != nil {
 				return err
 			}
 		}
+	}
 	return nil
 }
 
 // KEYS[1] -> source queue (e.g. scheduled or retry queue)
 // ARGV[1] -> current unix time
 // ARGV[2] -> queue prefix
+// Note: Script moves tasks up to 100 at a time to keep the runtime of script short.
 var forwardCmd = redis.NewScript(`
-local msgs = redis.call("ZRANGEBYSCORE", KEYS[1], "-inf", ARGV[1])
+local msgs = redis.call("ZRANGEBYSCORE", KEYS[1], "-inf", ARGV[1], "LIMIT", 0, 100)
 for _, msg in ipairs(msgs) do
 	local decoded = cjson.decode(msg)
 	local qkey = ARGV[2] .. decoded["Queue"]
 	redis.call("LPUSH", qkey, msg)
 	redis.call("ZREM", KEYS[1], msg)
 end
-return msgs`)
+return table.getn(msgs)`)
 
-// forward moves all tasks with a score less than the current unix time
-// from the src zset.
-func (r *RDB) forward(src string) error {
+// forward moves tasks with a score less than the current unix time
+// from the src zset. It returns the number of tasks moved.
+func (r *RDB) forward(src string) (int, error) {
 	now := float64(time.Now().Unix())
-	return forwardCmd.Run(r.client,
-		[]string{src}, now, base.QueuePrefix).Err()
-}
-
-// KEYS[1] -> source queue (e.g. scheduled or retry queue)
-// KEYS[2] -> destination queue
-var forwardSingleCmd = redis.NewScript(`
-local msgs = redis.call("ZRANGEBYSCORE", KEYS[1], "-inf", ARGV[1])
-for _, msg in ipairs(msgs) do
-	redis.call("LPUSH", KEYS[2], msg)
-	redis.call("ZREM", KEYS[1], msg)
-end
-return msgs`)
-
-// forwardSingle moves all tasks with a score less than the current unix time
-// from the src zset to dst list.
-func (r *RDB) forwardSingle(src, dst string) error {
-	now := float64(time.Now().Unix())
-	return forwardSingleCmd.Run(r.client,
-		[]string{src, dst}, now).Err()
+	res, err := forwardCmd.Run(r.client,
+		[]string{src}, now, base.QueuePrefix).Result()
+	if err != nil {
+		return 0, err
+	}
+	return cast.ToInt(res), nil
 }
 
 // KEYS[1]  -> asynq:servers:<host:pid:sid>
 // KEYS[2]  -> asynq:servers
 // KEYS[3]  -> asynq:workers<host:pid:sid>
-// keys[4]  -> asynq:workers
+// KEYS[4]  -> asynq:workers
 // ARGV[1]  -> expiration time
 // ARGV[2]  -> TTL in seconds
-// ARGV[3]  -> process info
+// ARGV[3]  -> server info
 // ARGV[4:] -> alternate key-value pair of (worker id, worker data)
 // Note: Add key to ZSET with expiration time as score.
 // ref: https://github.com/antirez/redis/issues/135#issuecomment-2361996
-var writeProcessInfoCmd = redis.NewScript(`
+var writeServerStateCmd = redis.NewScript(`
 redis.call("SETEX", KEYS[1], ARGV[2], ARGV[3])
 redis.call("ZADD", KEYS[2], ARGV[1], KEYS[1])
 redis.call("DEL", KEYS[3])
@@ -485,26 +438,23 @@ redis.call("ZADD", KEYS[4], ARGV[1], KEYS[3])
 return redis.status_reply("OK")`)
 
 // WriteServerState writes server state data to redis with expiration set to the value ttl.
-func (r *RDB) WriteServerState(ss *base.ServerState, ttl time.Duration) error {
-	info := ss.GetInfo()
+func (r *RDB) WriteServerState(info *base.ServerInfo, workers []*base.WorkerInfo, ttl time.Duration) error {
 	bytes, err := json.Marshal(info)
 	if err != nil {
 		return err
 	}
-	var args []interface{} // args to the lua script
 	exp := time.Now().Add(ttl).UTC()
-	workers := ss.GetWorkers()
-	args = append(args, float64(exp.Unix()), ttl.Seconds(), bytes)
+	args := []interface{}{float64(exp.Unix()), ttl.Seconds(), bytes} // args to the lua script
 	for _, w := range workers {
 		bytes, err := json.Marshal(w)
 		if err != nil {
 			continue // skip bad data
 		}
-		args = append(args, w.ID.String(), bytes)
+		args = append(args, w.ID, bytes)
 	}
 	skey := base.ServerInfoKey(info.Host, info.PID, info.ServerID)
 	wkey := base.WorkersKey(info.Host, info.PID, info.ServerID)
-	return writeProcessInfoCmd.Run(r.client,
+	return writeServerStateCmd.Run(r.client,
 		[]string{skey, base.AllServers, wkey, base.AllWorkers},
 		args...).Err()
 }
@@ -521,11 +471,9 @@ redis.call("DEL", KEYS[4])
 return redis.status_reply("OK")`)
 
 // ClearServerState deletes server state data from redis.
-func (r *RDB) ClearServerState(ss *base.ServerState) error {
-	info := ss.GetInfo()
-	host, pid, id := info.Host, info.PID, info.ServerID
-	skey := base.ServerInfoKey(host, pid, id)
-	wkey := base.WorkersKey(host, pid, id)
+func (r *RDB) ClearServerState(host string, pid int, serverID string) error {
+	skey := base.ServerInfoKey(host, pid, serverID)
+	wkey := base.WorkersKey(host, pid, serverID)
 	return clearProcessInfoCmd.Run(r.client,
 		[]string{base.AllServers, skey, base.AllWorkers, wkey}).Err()
 }

internal/rdb/rdb_test.go

@@ -227,6 +227,97 @@ func TestDequeue(t *testing.T) {
 	}
 }
 
+func TestDequeueIgnoresPausedQueues(t *testing.T) {
+	r := setup(t)
+	t1 := h.NewTaskMessage("send_email", map[string]interface{}{"subject": "hello!"})
+	t2 := h.NewTaskMessage("export_csv", nil)
+
+	tests := []struct {
+		paused         []string // list of paused queues
+		enqueued       map[string][]*base.TaskMessage
+		args           []string // list of queues to query
+		want           *base.TaskMessage
+		err            error
+		wantEnqueued   map[string][]*base.TaskMessage
+		wantInProgress []*base.TaskMessage
+	}{
+		{
+			paused: []string{"default"},
+			enqueued: map[string][]*base.TaskMessage{
+				"default":  {t1},
+				"critical": {t2},
+			},
+			args: []string{"default", "critical"},
+			want: t2,
+			err:  nil,
+			wantEnqueued: map[string][]*base.TaskMessage{
+				"default":  {t1},
+				"critical": {},
+			},
+			wantInProgress: []*base.TaskMessage{t2},
+		},
+		{
+			paused: []string{"default"},
+			enqueued: map[string][]*base.TaskMessage{
+				"default": {t1},
+			},
+			args: []string{"default"},
+			want: nil,
+			err:  ErrNoProcessableTask,
+			wantEnqueued: map[string][]*base.TaskMessage{
+				"default": {t1},
+			},
+			wantInProgress: []*base.TaskMessage{},
+		},
+		{
+			paused: []string{"critical", "default"},
+			enqueued: map[string][]*base.TaskMessage{
+				"default":  {t1},
+				"critical": {t2},
+			},
+			args: []string{"default", "critical"},
+			want: nil,
+			err:  ErrNoProcessableTask,
+			wantEnqueued: map[string][]*base.TaskMessage{
+				"default":  {t1},
+				"critical": {t2},
+			},
+			wantInProgress: []*base.TaskMessage{},
+		},
+	}
+
+	for _, tc := range tests {
+		h.FlushDB(t, r.client) // clean up db before each test case
+		for _, qname := range tc.paused {
+			if err := r.Pause(qname); err != nil {
+				t.Fatal(err)
+			}
+		}
+		for queue, msgs := range tc.enqueued {
+			h.SeedEnqueuedQueue(t, r.client, msgs, queue)
+		}
+
+		got, err := r.Dequeue(tc.args...)
+		if !cmp.Equal(got, tc.want) || err != tc.err {
+			t.Errorf("Dequeue(%v) = %v, %v; want %v, %v",
+				tc.args, got, err, tc.want, tc.err)
+			continue
+		}
+
+		for queue, want := range tc.wantEnqueued {
+			gotEnqueued := h.GetEnqueuedMessages(t, r.client, queue)
+			if diff := cmp.Diff(want, gotEnqueued, h.SortMsgOpt); diff != "" {
+				t.Errorf("mismatch found in %q: (-want,+got):\n%s", base.QueueKey(queue), diff)
+			}
+		}
+
+		gotInProgress := h.GetInProgressMessages(t, r.client)
+		if diff := cmp.Diff(tc.wantInProgress, gotInProgress, h.SortMsgOpt); diff != "" {
+			t.Errorf("mismatch found in %q: (-want,+got):\n%s", base.InProgressQueue, diff)
+		}
+	}
+}
+
 func TestDone(t *testing.T) {
 	r := setup(t)
 	t1 := h.NewTaskMessage("send_email", nil)
@@ -662,98 +753,6 @@ func TestKill(t *testing.T) {
 	}
 }
 
-func TestRequeueAll(t *testing.T) {
-	r := setup(t)
-	t1 := h.NewTaskMessage("send_email", nil)
-	t2 := h.NewTaskMessage("export_csv", nil)
-	t3 := h.NewTaskMessage("sync_stuff", nil)
-	t4 := h.NewTaskMessageWithQueue("important", nil, "critical")
-	t5 := h.NewTaskMessageWithQueue("minor", nil, "low")
-
-	tests := []struct {
-		inProgress     []*base.TaskMessage
-		enqueued       map[string][]*base.TaskMessage
-		want           int64
-		wantInProgress []*base.TaskMessage
-		wantEnqueued   map[string][]*base.TaskMessage
-	}{
-		{
-			inProgress: []*base.TaskMessage{t1, t2, t3},
-			enqueued: map[string][]*base.TaskMessage{
-				base.DefaultQueueName: {},
-			},
-			want:           3,
-			wantInProgress: []*base.TaskMessage{},
-			wantEnqueued: map[string][]*base.TaskMessage{
-				base.DefaultQueueName: {t1, t2, t3},
-			},
-		},
-		{
-			inProgress: []*base.TaskMessage{},
-			enqueued: map[string][]*base.TaskMessage{
-				base.DefaultQueueName: {t1, t2, t3},
-			},
-			want:           0,
-			wantInProgress: []*base.TaskMessage{},
-			wantEnqueued: map[string][]*base.TaskMessage{
-				base.DefaultQueueName: {t1, t2, t3},
-			},
-		},
-		{
-			inProgress: []*base.TaskMessage{t2, t3},
-			enqueued: map[string][]*base.TaskMessage{
-				base.DefaultQueueName: {t1},
-			},
-			want:           2,
-			wantInProgress: []*base.TaskMessage{},
-			wantEnqueued: map[string][]*base.TaskMessage{
-				base.DefaultQueueName: {t1, t2, t3},
-			},
-		},
-		{
-			inProgress: []*base.TaskMessage{t2, t3, t4, t5},
-			enqueued: map[string][]*base.TaskMessage{
-				base.DefaultQueueName: {t1},
-				"critical":            {},
-				"low":                 {},
-			},
-			want:           4,
-			wantInProgress: []*base.TaskMessage{},
-			wantEnqueued: map[string][]*base.TaskMessage{
-				base.DefaultQueueName: {t1, t2, t3},
-				"critical":            {t4},
-				"low":                 {t5},
-			},
-		},
-	}
-
-	for _, tc := range tests {
-		h.FlushDB(t, r.client) // clean up db before each test case
-		h.SeedInProgressQueue(t, r.client, tc.inProgress)
-		for qname, msgs := range tc.enqueued {
-			h.SeedEnqueuedQueue(t, r.client, msgs, qname)
-		}
-
-		got, err := r.RequeueAll()
-		if got != tc.want || err != nil {
-			t.Errorf("(*RDB).RequeueAll() = %v %v, want %v nil", got, err, tc.want)
-			continue
-		}
-
-		gotInProgress := h.GetInProgressMessages(t, r.client)
-		if diff := cmp.Diff(tc.wantInProgress, gotInProgress, h.SortMsgOpt); diff != "" {
-			t.Errorf("mismatch found in %q: (-want, +got):\n%s", base.InProgressQueue, diff)
-		}
-
-		for qname, want := range tc.wantEnqueued {
-			gotEnqueued := h.GetEnqueuedMessages(t, r.client, qname)
-			if diff := cmp.Diff(want, gotEnqueued, h.SortMsgOpt); diff != "" {
-				t.Errorf("mismatch found in %q: (-want, +got):\n%s", base.QueueKey(qname), diff)
-			}
-		}
-	}
-}
-
 func TestCheckAndEnqueue(t *testing.T) {
 	r := setup(t)
 	t1 := h.NewTaskMessage("send_email", nil)
@@ -769,7 +768,6 @@ func TestCheckAndEnqueue(t *testing.T) {
 	tests := []struct {
 		scheduled     []h.ZSetEntry
 		retry         []h.ZSetEntry
-		qnames        []string
 		wantEnqueued  map[string][]*base.TaskMessage
 		wantScheduled []*base.TaskMessage
 		wantRetry     []*base.TaskMessage
@@ -781,7 +779,6 @@ func TestCheckAndEnqueue(t *testing.T) {
 			},
 			retry: []h.ZSetEntry{
 				{Msg: t3, Score: float64(secondAgo.Unix())}},
-			qnames: []string{"default"},
 			wantEnqueued: map[string][]*base.TaskMessage{
 				"default": {t1, t2, t3},
 			},
@@ -794,7 +791,6 @@ func TestCheckAndEnqueue(t *testing.T) {
 				{Msg: t2, Score: float64(secondAgo.Unix())}},
 			retry: []h.ZSetEntry{
 				{Msg: t3, Score: float64(secondAgo.Unix())}},
-			qnames: []string{"default"},
 			wantEnqueued: map[string][]*base.TaskMessage{
 				"default": {t2, t3},
 			},
@@ -807,7 +803,6 @@ func TestCheckAndEnqueue(t *testing.T) {
 				{Msg: t2, Score: float64(hourFromNow.Unix())}},
 			retry: []h.ZSetEntry{
 				{Msg: t3, Score: float64(hourFromNow.Unix())}},
-			qnames: []string{"default"},
 			wantEnqueued: map[string][]*base.TaskMessage{
 				"default": {},
 			},
@@ -821,7 +816,6 @@ func TestCheckAndEnqueue(t *testing.T) {
 			},
 			retry: []h.ZSetEntry{
 				{Msg: t5, Score: float64(secondAgo.Unix())}},
-			qnames: []string{"default", "critical", "low"},
 			wantEnqueued: map[string][]*base.TaskMessage{
 				"default":  {t1},
 				"critical": {t4},
@@ -837,7 +831,7 @@ func TestCheckAndEnqueue(t *testing.T) {
 		h.SeedScheduledQueue(t, r.client, tc.scheduled)
 		h.SeedRetryQueue(t, r.client, tc.retry)
 
-		err := r.CheckAndEnqueue(tc.qnames...)
+		err := r.CheckAndEnqueue()
 		if err != nil {
 			t.Errorf("(*RDB).CheckScheduled() = %v, want nil", err)
 			continue
@@ -864,64 +858,63 @@ func TestCheckAndEnqueue(t *testing.T) {
 
 func TestWriteServerState(t *testing.T) {
 	r := setup(t)
-	queues := map[string]int{"default": 2, "email": 5, "low": 1}
 
-	started := time.Now()
-	ss := base.NewServerState("localhost", 4242, 10, queues, false)
-	ss.SetStarted(started)
-	ss.SetStatus(base.StatusRunning)
-	ttl := 5 * time.Second
+	var (
+		host     = "localhost"
+		pid      = 4242
+		serverID = "server123"
 
-	h.FlushDB(t, r.client)
+		ttl = 5 * time.Second
+	)
 
-	err := r.WriteServerState(ss, ttl)
+	info := base.ServerInfo{
+		Host:              host,
+		PID:               pid,
+		ServerID:          serverID,
+		Concurrency:       10,
+		Queues:            map[string]int{"default": 2, "email": 5, "low": 1},
+		StrictPriority:    false,
+		Started:           time.Now(),
+		Status:            "running",
+		ActiveWorkerCount: 0,
+	}
+
+	err := r.WriteServerState(&info, nil /* workers */, ttl)
 	if err != nil {
 		t.Errorf("r.WriteServerState returned an error: %v", err)
 	}
 
-	// Check ServerInfo was written correctly
-	info := ss.GetInfo()
-	skey := base.ServerInfoKey(info.Host, info.PID, info.ServerID)
+	// Check ServerInfo was written correctly.
+	skey := base.ServerInfoKey(host, pid, serverID)
 	data := r.client.Get(skey).Val()
 	var got base.ServerInfo
 	err = json.Unmarshal([]byte(data), &got)
 	if err != nil {
 		t.Fatalf("could not decode json: %v", err)
 	}
-	want := base.ServerInfo{
-		Host:              info.Host,
-		PID:               info.PID,
-		Concurrency:       info.Concurrency,
-		Queues:            map[string]int{"default": 2, "email": 5, "low": 1},
-		StrictPriority:    false,
-		Status:            "running",
-		Started:           started,
-		ActiveWorkerCount: 0,
-	}
-	ignoreOpt := cmpopts.IgnoreFields(base.ServerInfo{}, "ServerID")
-	if diff := cmp.Diff(want, got, ignoreOpt); diff != "" {
+	if diff := cmp.Diff(info, got); diff != "" {
 		t.Errorf("persisted ServerInfo was %v, want %v; (-want,+got)\n%s",
-			got, want, diff)
+			got, info, diff)
 	}
-	// Check ServerInfo TTL was set correctly
+	// Check ServerInfo TTL was set correctly.
 	gotTTL := r.client.TTL(skey).Val()
 	if !cmp.Equal(ttl.Seconds(), gotTTL.Seconds(), cmpopts.EquateApprox(0, 1)) {
 		t.Errorf("TTL of %q was %v, want %v", skey, gotTTL, ttl)
 	}
-	// Check ServerInfo key was added to the set correctly
-	gotProcesses := r.client.ZRange(base.AllServers, 0, -1).Val()
-	wantProcesses := []string{skey}
-	if diff := cmp.Diff(wantProcesses, gotProcesses); diff != "" {
-		t.Errorf("%q contained %v, want %v", base.AllServers, gotProcesses, wantProcesses)
+	// Check ServerInfo key was added to the set all server keys correctly.
+	gotServerKeys := r.client.ZRange(base.AllServers, 0, -1).Val()
+	wantServerKeys := []string{skey}
+	if diff := cmp.Diff(wantServerKeys, gotServerKeys); diff != "" {
+		t.Errorf("%q contained %v, want %v", base.AllServers, gotServerKeys, wantServerKeys)
 	}
 
-	// Check WorkersInfo was written correctly
-	wkey := base.WorkersKey(info.Host, info.PID, info.ServerID)
+	// Check WorkersInfo was written correctly.
+	wkey := base.WorkersKey(host, pid, serverID)
 	workerExist := r.client.Exists(wkey).Val()
 	if workerExist != 0 {
 		t.Errorf("%q key exists", wkey)
 	}
-	// Check WorkersInfo key was added to the set correctly
+	// Check WorkersInfo key was added to the set correctly.
 	gotWorkerKeys := r.client.ZRange(base.AllWorkers, 0, -1).Val()
 	wantWorkerKeys := []string{wkey}
 	if diff := cmp.Diff(wantWorkerKeys, gotWorkerKeys); diff != "" {
@@ -931,109 +924,105 @@ func TestWriteServerState(t *testing.T) {
 
 func TestWriteServerStateWithWorkers(t *testing.T) {
 	r := setup(t)
-	queues := map[string]int{"default": 2, "email": 5, "low": 1}
-	concurrency := 10
 
-	started := time.Now().Add(-10 * time.Minute)
-	w1Started := time.Now().Add(-time.Minute)
-	w2Started := time.Now().Add(-time.Second)
-	msg1 := h.NewTaskMessage("send_email", map[string]interface{}{"user_id": "123"})
-	msg2 := h.NewTaskMessage("gen_thumbnail", map[string]interface{}{"path": "some/path/to/imgfile"})
-	ss := base.NewServerState("127.0.01", 4242, concurrency, queues, false)
-	ss.SetStarted(started)
-	ss.SetStatus(base.StatusRunning)
-	ss.AddWorkerStats(msg1, w1Started)
-	ss.AddWorkerStats(msg2, w2Started)
-	ttl := 5 * time.Second
+	var (
+		host     = "127.0.0.1"
+		pid      = 4242
+		serverID = "server123"
 
-	h.FlushDB(t, r.client)
+		msg1 = h.NewTaskMessage("send_email", map[string]interface{}{"user_id": "123"})
+		msg2 = h.NewTaskMessage("gen_thumbnail", map[string]interface{}{"path": "some/path/to/imgfile"})
 
-	err := r.WriteServerState(ss, ttl)
-	if err != nil {
-		t.Errorf("r.WriteServerState returned an error: %v", err)
+		ttl = 5 * time.Second
+	)
+
+	workers := []*base.WorkerInfo{
+		{
+			Host:    host,
+			PID:     pid,
+			ID:      msg1.ID.String(),
+			Type:    msg1.Type,
+			Queue:   msg1.Queue,
+			Payload: msg1.Payload,
+			Started: time.Now().Add(-10 * time.Second),
+		},
+		{
+			Host:    host,
+			PID:     pid,
+			ID:      msg2.ID.String(),
+			Type:    msg2.Type,
+			Queue:   msg2.Queue,
+			Payload: msg2.Payload,
+			Started: time.Now().Add(-2 * time.Minute),
+		},
 	}
 
-	// Check ServerInfo was written correctly
-	info := ss.GetInfo()
-	skey := base.ServerInfoKey(info.Host, info.PID, info.ServerID)
+	serverInfo := base.ServerInfo{
+		Host:              host,
+		PID:               pid,
+		ServerID:          serverID,
+		Concurrency:       10,
+		Queues:            map[string]int{"default": 2, "email": 5, "low": 1},
+		StrictPriority:    false,
+		Started:           time.Now().Add(-10 * time.Minute),
+		Status:            "running",
+		ActiveWorkerCount: len(workers),
+	}
+
+	err := r.WriteServerState(&serverInfo, workers, ttl)
+	if err != nil {
+		t.Fatalf("r.WriteServerState returned an error: %v", err)
+	}
+
+	// Check ServerInfo was written correctly.
+	skey := base.ServerInfoKey(host, pid, serverID)
 	data := r.client.Get(skey).Val()
 	var got base.ServerInfo
 	err = json.Unmarshal([]byte(data), &got)
 	if err != nil {
 		t.Fatalf("could not decode json: %v", err)
 	}
-	want := base.ServerInfo{
-		Host:              info.Host,
-		PID:               info.PID,
-		ServerID:          info.ServerID,
-		Concurrency:       concurrency,
-		Queues:            queues,
-		StrictPriority:    false,
-		Status:            "running",
-		Started:           started,
-		ActiveWorkerCount: 2,
-	}
-	if diff := cmp.Diff(want, got); diff != "" {
+	if diff := cmp.Diff(serverInfo, got); diff != "" {
 		t.Errorf("persisted ServerInfo was %v, want %v; (-want,+got)\n%s",
-			got, want, diff)
+			got, serverInfo, diff)
 	}
-	// Check ServerInfo TTL was set correctly
+	// Check ServerInfo TTL was set correctly.
 	gotTTL := r.client.TTL(skey).Val()
 	if !cmp.Equal(ttl.Seconds(), gotTTL.Seconds(), cmpopts.EquateApprox(0, 1)) {
 		t.Errorf("TTL of %q was %v, want %v", skey, gotTTL, ttl)
 	}
-	// Check ServerInfo key was added to the set correctly
-	gotProcesses := r.client.ZRange(base.AllServers, 0, -1).Val()
-	wantProcesses := []string{skey}
-	if diff := cmp.Diff(wantProcesses, gotProcesses); diff != "" {
-		t.Errorf("%q contained %v, want %v", base.AllServers, gotProcesses, wantProcesses)
+	// Check ServerInfo key was added to the set correctly.
+	gotServerKeys := r.client.ZRange(base.AllServers, 0, -1).Val()
+	wantServerKeys := []string{skey}
+	if diff := cmp.Diff(wantServerKeys, gotServerKeys); diff != "" {
+		t.Errorf("%q contained %v, want %v", base.AllServers, gotServerKeys, wantServerKeys)
 	}
 
-	// Check WorkersInfo was written correctly
-	wkey := base.WorkersKey(info.Host, info.PID, info.ServerID)
+	// Check WorkersInfo was written correctly.
+	wkey := base.WorkersKey(host, pid, serverID)
 	wdata := r.client.HGetAll(wkey).Val()
 	if len(wdata) != 2 {
 		t.Fatalf("HGETALL %q returned a hash of size %d, want 2", wkey, len(wdata))
 	}
-	gotWorkers := make(map[string]*base.WorkerInfo)
-	for key, val := range wdata {
+	var gotWorkers []*base.WorkerInfo
+	for _, val := range wdata {
 		var w base.WorkerInfo
 		if err := json.Unmarshal([]byte(val), &w); err != nil {
 			t.Fatalf("could not unmarshal worker's data: %v", err)
 		}
-		gotWorkers[key] = &w
+		gotWorkers = append(gotWorkers, &w)
 	}
-	wantWorkers := map[string]*base.WorkerInfo{
-		msg1.ID.String(): {
-			Host:    info.Host,
-			PID:     info.PID,
-			ID:      msg1.ID,
-			Type:    msg1.Type,
-			Queue:   msg1.Queue,
-			Payload: msg1.Payload,
-			Started: w1Started,
-		},
-		msg2.ID.String(): {
-			Host:    info.Host,
-			PID:     info.PID,
-			ID:      msg2.ID,
-			Type:    msg2.Type,
-			Queue:   msg2.Queue,
-			Payload: msg2.Payload,
-			Started: w2Started,
-		},
-	}
-	if diff := cmp.Diff(wantWorkers, gotWorkers); diff != "" {
+	if diff := cmp.Diff(workers, gotWorkers, h.SortWorkerInfoOpt); diff != "" {
 		t.Errorf("persisted workers info was %v, want %v; (-want,+got)\n%s",
-			gotWorkers, wantWorkers, diff)
+			gotWorkers, workers, diff)
 	}
 
-	// Check WorkersInfo TTL was set correctly
+	// Check WorkersInfo TTL was set correctly.
 	gotTTL = r.client.TTL(wkey).Val()
-	if !cmp.Equal(ttl, gotTTL, timeCmpOpt) {
+	if !cmp.Equal(ttl.Seconds(), gotTTL.Seconds(), cmpopts.EquateApprox(0, 1)) {
 		t.Errorf("TTL of %q was %v, want %v", wkey, gotTTL, ttl)
 	}
-	// Check WorkersInfo key was added to the set correctly
+	// Check WorkersInfo key was added to the set correctly.
 	gotWorkerKeys := r.client.ZRange(base.AllWorkers, 0, -1).Val()
 	wantWorkerKeys := []string{wkey}
 	if diff := cmp.Diff(wantWorkerKeys, gotWorkerKeys); diff != "" {
@@ -1043,51 +1032,96 @@ func TestWriteServerStateWithWorkers(t *testing.T) {
 
 func TestClearServerState(t *testing.T) {
 	r := setup(t)
-	ss := base.NewServerState("127.0.01", 4242, 10, map[string]int{"default": 1}, false)
-	info := ss.GetInfo()
 
-	h.FlushDB(t, r.client)
+	var (
+		host     = "127.0.0.1"
+		pid      = 1234
+		serverID = "server123"
 
-	skey := base.ServerInfoKey(info.Host, info.PID, info.ServerID)
-	wkey := base.WorkersKey(info.Host, info.PID, info.ServerID)
-	otherSKey := base.ServerInfoKey("otherhost", 12345, "server98")
-	otherWKey := base.WorkersKey("otherhost", 12345, "server98")
-	// Populate the keys.
-	if err := r.client.Set(skey, "process-info", 0).Err(); err != nil {
-		t.Fatal(err)
+		otherHost     = "127.0.0.2"
+		otherPID      = 9876
+		otherServerID = "server987"
+
+		msg1 = h.NewTaskMessage("send_email", map[string]interface{}{"user_id": "123"})
+		msg2 = h.NewTaskMessage("gen_thumbnail", map[string]interface{}{"path": "some/path/to/imgfile"})
+
+		ttl = 5 * time.Second
+	)
+
+	workers1 := []*base.WorkerInfo{
+		{
+			Host:    host,
+			PID:     pid,
+			ID:      msg1.ID.String(),
+			Type:    msg1.Type,
+			Queue:   msg1.Queue,
+			Payload: msg1.Payload,
+			Started: time.Now().Add(-10 * time.Second),
+		},
 	}
-	if err := r.client.HSet(wkey, "worker-key", "worker-info").Err(); err != nil {
-		t.Fatal(err)
-	}
-	if err := r.client.ZAdd(base.AllServers, &redis.Z{Member: skey}).Err(); err != nil {
-		t.Fatal(err)
-	}
-	if err := r.client.ZAdd(base.AllServers, &redis.Z{Member: otherSKey}).Err(); err != nil {
-		t.Fatal(err)
-	}
-	if err := r.client.ZAdd(base.AllWorkers, &redis.Z{Member: wkey}).Err(); err != nil {
-		t.Fatal(err)
-	}
-	if err := r.client.ZAdd(base.AllWorkers, &redis.Z{Member: otherWKey}).Err(); err != nil {
-		t.Fatal(err)
+	serverInfo1 := base.ServerInfo{
+		Host:              host,
+		PID:               pid,
+		ServerID:          serverID,
+		Concurrency:       10,
+		Queues:            map[string]int{"default": 2, "email": 5, "low": 1},
+		StrictPriority:    false,
+		Started:           time.Now().Add(-10 * time.Minute),
+		Status:            "running",
+		ActiveWorkerCount: len(workers1),
 	}
 
-	err := r.ClearServerState(ss)
+	workers2 := []*base.WorkerInfo{
+		{
+			Host:    otherHost,
+			PID:     otherPID,
+			ID:      msg2.ID.String(),
+			Type:    msg2.Type,
+			Queue:   msg2.Queue,
+			Payload: msg2.Payload,
+			Started: time.Now().Add(-30 * time.Second),
+		},
+	}
+	serverInfo2 := base.ServerInfo{
+		Host:              otherHost,
+		PID:               otherPID,
+		ServerID:          otherServerID,
+		Concurrency:       10,
+		Queues:            map[string]int{"default": 2, "email": 5, "low": 1},
+		StrictPriority:    false,
+		Started:           time.Now().Add(-15 * time.Minute),
+		Status:            "running",
+		ActiveWorkerCount: len(workers2),
+	}
+
+	// Write server and workers data.
+	if err := r.WriteServerState(&serverInfo1, workers1, ttl); err != nil {
+		t.Fatalf("could not write server state: %v", err)
+	}
+	if err := r.WriteServerState(&serverInfo2, workers2, ttl); err != nil {
+		t.Fatalf("could not write server state: %v", err)
+	}
+
+	err := r.ClearServerState(host, pid, serverID)
 	if err != nil {
 		t.Fatalf("(*RDB).ClearServerState failed: %v", err)
 	}
 
-	// Check all keys are cleared
+	skey := base.ServerInfoKey(host, pid, serverID)
+	wkey := base.WorkersKey(host, pid, serverID)
+	otherSKey := base.ServerInfoKey(otherHost, otherPID, otherServerID)
+	otherWKey := base.WorkersKey(otherHost, otherPID, otherServerID)
+	// Check all keys are cleared.
 	if r.client.Exists(skey).Val() != 0 {
 		t.Errorf("Redis key %q exists", skey)
 	}
 	if r.client.Exists(wkey).Val() != 0 {
 		t.Errorf("Redis key %q exists", wkey)
 	}
-	gotProcessKeys := r.client.ZRange(base.AllServers, 0, -1).Val()
-	wantProcessKeys := []string{otherSKey}
-	if diff := cmp.Diff(wantProcessKeys, gotProcessKeys); diff != "" {
-		t.Errorf("%q contained %v, want %v", base.AllServers, gotProcessKeys, wantProcessKeys)
+	gotServerKeys := r.client.ZRange(base.AllServers, 0, -1).Val()
+	wantServerKeys := []string{otherSKey}
+	if diff := cmp.Diff(wantServerKeys, gotServerKeys); diff != "" {
+		t.Errorf("%q contained %v, want %v", base.AllServers, gotServerKeys, wantServerKeys)
 	}
 	gotWorkerKeys := r.client.ZRange(base.AllWorkers, 0, -1).Val()
 	wantWorkerKeys := []string{otherWKey}

internal/testbroker/testbroker.go

@@ -123,16 +123,7 @@ func (tb *TestBroker) Kill(msg *base.TaskMessage, errMsg string) error {
 	return tb.real.Kill(msg, errMsg)
 }
 
-func (tb *TestBroker) RequeueAll() (int64, error) {
-	tb.mu.Lock()
-	defer tb.mu.Unlock()
-	if tb.sleeping {
-		return 0, errRedisDown
-	}
-	return tb.real.RequeueAll()
-}
-
-func (tb *TestBroker) CheckAndEnqueue(qnames ...string) error {
+func (tb *TestBroker) CheckAndEnqueue() error {
 	tb.mu.Lock()
 	defer tb.mu.Unlock()
 	if tb.sleeping {
@@ -141,22 +132,22 @@ func (tb *TestBroker) CheckAndEnqueue(qnames ...string) error {
 	return tb.real.CheckAndEnqueue()
 }
 
-func (tb *TestBroker) WriteServerState(ss *base.ServerState, ttl time.Duration) error {
+func (tb *TestBroker) WriteServerState(info *base.ServerInfo, workers []*base.WorkerInfo, ttl time.Duration) error {
 	tb.mu.Lock()
 	defer tb.mu.Unlock()
 	if tb.sleeping {
 		return errRedisDown
 	}
-	return tb.real.WriteServerState(ss, ttl)
+	return tb.real.WriteServerState(info, workers, ttl)
 }
 
-func (tb *TestBroker) ClearServerState(ss *base.ServerState) error {
+func (tb *TestBroker) ClearServerState(host string, pid int, serverID string) error {
 	tb.mu.Lock()
 	defer tb.mu.Unlock()
 	if tb.sleeping {
 		return errRedisDown
 	}
-	return tb.real.ClearServerState(ss)
+	return tb.real.ClearServerState(host, pid, serverID)
 }
 
 func (tb *TestBroker) CancelationPubSub() (*redis.PubSub, error) {

payload.go

@@ -5,6 +5,7 @@
 package asynq
 
 import (
+	"encoding/json"
 	"fmt"
 	"time"
 
@@ -30,6 +31,19 @@ func (p Payload) Has(key string) bool {
 	return ok
 }
 
+func toInt(v interface{}) (int, error) {
+	switch v := v.(type) {
+	case json.Number:
+		val, err := v.Int64()
+		if err != nil {
+			return 0, err
+		}
+		return int(val), nil
+	default:
+		return cast.ToIntE(v)
+	}
+}
+
 // GetString returns a string value if a string type is associated with
 // the key, otherwise reports an error.
 func (p Payload) GetString(key string) (string, error) {
@@ -47,7 +61,7 @@ func (p Payload) GetInt(key string) (int, error) {
 	if !ok {
 		return 0, &errKeyNotFound{key}
 	}
-	return cast.ToIntE(v)
+	return toInt(v)
 }
 
 // GetFloat64 returns a float64 value if a numeric type is associated with
@@ -57,7 +71,12 @@ func (p Payload) GetFloat64(key string) (float64, error) {
 	if !ok {
 		return 0, &errKeyNotFound{key}
 	}
+	switch v := v.(type) {
+	case json.Number:
+		return v.Float64()
+	default:
 		return cast.ToFloat64E(v)
+	}
 }
 
 // GetBool returns a boolean value if a boolean type is associated with
@@ -87,7 +106,20 @@ func (p Payload) GetIntSlice(key string) ([]int, error) {
 	if !ok {
 		return nil, &errKeyNotFound{key}
 	}
+	switch v := v.(type) {
+	case []interface{}:
+		var res []int
+		for _, elem := range v {
+			val, err := toInt(elem)
+			if err != nil {
+				return nil, err
+			}
+			res = append(res, int(val))
+		}
+		return res, nil
+	default:
 		return cast.ToIntSliceE(v)
+	}
 }
 
 // GetStringMap returns a map of string to empty interface
@@ -131,7 +163,20 @@ func (p Payload) GetStringMapInt(key string) (map[string]int, error) {
 	if !ok {
 		return nil, &errKeyNotFound{key}
 	}
+	switch v := v.(type) {
+	case map[string]interface{}:
+		res := make(map[string]int)
+		for key, val := range v {
+			ival, err := toInt(val)
+			if err != nil {
+				return nil, err
+			}
+			res[key] = ival
+		}
+		return res, nil
+	default:
 		return cast.ToStringMapIntE(v)
+	}
 }
 
 // GetStringMapBool returns a map of string to boolean
@@ -162,5 +207,14 @@ func (p Payload) GetDuration(key string) (time.Duration, error) {
 	if !ok {
 		return 0, &errKeyNotFound{key}
 	}
+	switch v := v.(type) {
+	case json.Number:
+		val, err := v.Int64()
+		if err != nil {
+			return 0, err
+		}
+		return time.Duration(val), nil
+	default:
 		return cast.ToDurationE(v)
+	}
 }

payload_test.go

@@ -10,6 +10,7 @@ import (
 	"time"
 
 	"github.com/google/go-cmp/cmp"
+	"github.com/google/go-cmp/cmp/cmpopts"
 	h "github.com/hibiken/asynq/internal/asynqtest"
 	"github.com/hibiken/asynq/internal/base"
 )
@@ -40,12 +41,11 @@ func TestPayloadString(t *testing.T) {
 
 		// encode and then decode task messsage.
 		in := h.NewTaskMessage("testing", tc.data)
-		b, err := json.Marshal(in)
+		encoded, err := base.EncodeMessage(in)
 		if err != nil {
 			t.Fatal(err)
 		}
-		var out base.TaskMessage
-		err = json.Unmarshal(b, &out)
+		out, err := base.DecodeMessage(encoded)
 		if err != nil {
 			t.Fatal(err)
 		}
@@ -85,12 +85,11 @@ func TestPayloadInt(t *testing.T) {
 
 		// encode and then decode task messsage.
 		in := h.NewTaskMessage("testing", tc.data)
-		b, err := json.Marshal(in)
+		encoded, err := base.EncodeMessage(in)
 		if err != nil {
 			t.Fatal(err)
 		}
-		var out base.TaskMessage
-		err = json.Unmarshal(b, &out)
+		out, err := base.DecodeMessage(encoded)
 		if err != nil {
 			t.Fatal(err)
 		}
@@ -130,12 +129,11 @@ func TestPayloadFloat64(t *testing.T) {
 
 		// encode and then decode task messsage.
 		in := h.NewTaskMessage("testing", tc.data)
-		b, err := json.Marshal(in)
+		encoded, err := base.EncodeMessage(in)
 		if err != nil {
 			t.Fatal(err)
 		}
-		var out base.TaskMessage
-		err = json.Unmarshal(b, &out)
+		out, err := base.DecodeMessage(encoded)
 		if err != nil {
 			t.Fatal(err)
 		}
@@ -175,12 +173,11 @@ func TestPayloadBool(t *testing.T) {
 
 		// encode and then decode task messsage.
 		in := h.NewTaskMessage("testing", tc.data)
-		b, err := json.Marshal(in)
+		encoded, err := base.EncodeMessage(in)
 		if err != nil {
 			t.Fatal(err)
 		}
-		var out base.TaskMessage
-		err = json.Unmarshal(b, &out)
+		out, err := base.DecodeMessage(encoded)
 		if err != nil {
 			t.Fatal(err)
 		}
@@ -221,12 +218,11 @@ func TestPayloadStringSlice(t *testing.T) {
 
 		// encode and then decode task messsage.
 		in := h.NewTaskMessage("testing", tc.data)
-		b, err := json.Marshal(in)
+		encoded, err := base.EncodeMessage(in)
 		if err != nil {
 			t.Fatal(err)
 		}
-		var out base.TaskMessage
-		err = json.Unmarshal(b, &out)
+		out, err := base.DecodeMessage(encoded)
 		if err != nil {
 			t.Fatal(err)
 		}
@@ -268,12 +264,11 @@ func TestPayloadIntSlice(t *testing.T) {
 
 		// encode and then decode task messsage.
 		in := h.NewTaskMessage("testing", tc.data)
-		b, err := json.Marshal(in)
+		encoded, err := base.EncodeMessage(in)
 		if err != nil {
 			t.Fatal(err)
 		}
-		var out base.TaskMessage
-		err = json.Unmarshal(b, &out)
+		out, err := base.DecodeMessage(encoded)
 		if err != nil {
 			t.Fatal(err)
 		}
@@ -315,21 +310,28 @@ func TestPayloadStringMap(t *testing.T) {
 
 		// encode and then decode task messsage.
 		in := h.NewTaskMessage("testing", tc.data)
-		b, err := json.Marshal(in)
+		encoded, err := base.EncodeMessage(in)
 		if err != nil {
 			t.Fatal(err)
 		}
-		var out base.TaskMessage
-		err = json.Unmarshal(b, &out)
+		out, err := base.DecodeMessage(encoded)
 		if err != nil {
 			t.Fatal(err)
 		}
 		payload = Payload{out.Payload}
 		got, err = payload.GetStringMap(tc.key)
-		diff = cmp.Diff(got, tc.data[tc.key])
+		ignoreOpt := cmpopts.IgnoreMapEntries(func(key string, val interface{}) bool {
+			switch val.(type) {
+			case json.Number:
+				return true
+			default:
+				return false
+			}
+		})
+		diff = cmp.Diff(got, tc.data[tc.key], ignoreOpt)
 		if err != nil || diff != "" {
-			t.Errorf("With Marshaling: Payload.GetStringMap(%q) = %v, %v, want %v, nil",
-				tc.key, got, err, tc.data[tc.key])
+			t.Errorf("With Marshaling: Payload.GetStringMap(%q) = %v, %v, want %v, nil;(-want,+got)\n%s",
+				tc.key, got, err, tc.data[tc.key], diff)
 		}
 
 		// access non-existent key.
@@ -362,12 +364,11 @@ func TestPayloadStringMapString(t *testing.T) {
 
 		// encode and then decode task messsage.
 		in := h.NewTaskMessage("testing", tc.data)
-		b, err := json.Marshal(in)
+		encoded, err := base.EncodeMessage(in)
 		if err != nil {
 			t.Fatal(err)
 		}
-		var out base.TaskMessage
-		err = json.Unmarshal(b, &out)
+		out, err := base.DecodeMessage(encoded)
 		if err != nil {
 			t.Fatal(err)
 		}
@@ -413,12 +414,11 @@ func TestPayloadStringMapStringSlice(t *testing.T) {
 
 		// encode and then decode task messsage.
 		in := h.NewTaskMessage("testing", tc.data)
-		b, err := json.Marshal(in)
+		encoded, err := base.EncodeMessage(in)
 		if err != nil {
 			t.Fatal(err)
 		}
-		var out base.TaskMessage
-		err = json.Unmarshal(b, &out)
+		out, err := base.DecodeMessage(encoded)
 		if err != nil {
 			t.Fatal(err)
 		}
@@ -465,12 +465,11 @@ func TestPayloadStringMapInt(t *testing.T) {
 
 		// encode and then decode task messsage.
 		in := h.NewTaskMessage("testing", tc.data)
-		b, err := json.Marshal(in)
+		encoded, err := base.EncodeMessage(in)
 		if err != nil {
 			t.Fatal(err)
 		}
-		var out base.TaskMessage
-		err = json.Unmarshal(b, &out)
+		out, err := base.DecodeMessage(encoded)
 		if err != nil {
 			t.Fatal(err)
 		}
@@ -517,12 +516,11 @@ func TestPayloadStringMapBool(t *testing.T) {
 
 		// encode and then decode task messsage.
 		in := h.NewTaskMessage("testing", tc.data)
-		b, err := json.Marshal(in)
+		encoded, err := base.EncodeMessage(in)
 		if err != nil {
 			t.Fatal(err)
 		}
-		var out base.TaskMessage
-		err = json.Unmarshal(b, &out)
+		out, err := base.DecodeMessage(encoded)
 		if err != nil {
 			t.Fatal(err)
 		}
@@ -564,12 +562,11 @@ func TestPayloadTime(t *testing.T) {
 
 		// encode and then decode task messsage.
 		in := h.NewTaskMessage("testing", tc.data)
-		b, err := json.Marshal(in)
+		encoded, err := base.EncodeMessage(in)
 		if err != nil {
 			t.Fatal(err)
 		}
-		var out base.TaskMessage
-		err = json.Unmarshal(b, &out)
+		out, err := base.DecodeMessage(encoded)
 		if err != nil {
 			t.Fatal(err)
 		}
@@ -611,12 +608,11 @@ func TestPayloadDuration(t *testing.T) {
 
 		// encode and then decode task messsage.
 		in := h.NewTaskMessage("testing", tc.data)
-		b, err := json.Marshal(in)
+		encoded, err := base.EncodeMessage(in)
 		if err != nil {
 			t.Fatal(err)
 		}
-		var out base.TaskMessage
-		err = json.Unmarshal(b, &out)
+		out, err := base.DecodeMessage(encoded)
 		if err != nil {
 			t.Fatal(err)
 		}

processor.go

@@ -22,8 +22,6 @@ type processor struct {
 	logger *log.Logger
 	broker base.Broker
 
-	ss *base.ServerState
-
 	handler Handler
 
 	queueConfig map[string]int
@@ -60,6 +58,9 @@ type processor struct {
 
 	// cancelations is a set of cancel functions for all in-progress tasks.
 	cancelations *base.Cancelations
+
+	starting chan<- *base.TaskMessage
+	finished chan<- *base.TaskMessage
 }
 
 type retryDelayFunc func(n int, err error, task *Task) time.Duration
@@ -67,38 +68,42 @@ type retryDelayFunc func(n int, err error, task *Task) time.Duration
 type processorParams struct {
 	logger          *log.Logger
 	broker          base.Broker
-	ss              *base.ServerState
 	retryDelayFunc  retryDelayFunc
 	syncCh          chan<- *syncRequest
 	cancelations    *base.Cancelations
+	concurrency     int
+	queues          map[string]int
+	strictPriority  bool
 	errHandler      ErrorHandler
 	shutdownTimeout time.Duration
+	starting        chan<- *base.TaskMessage
+	finished        chan<- *base.TaskMessage
 }
 
 // newProcessor constructs a new processor.
 func newProcessor(params processorParams) *processor {
-	info := params.ss.GetInfo()
-	qcfg := normalizeQueueCfg(info.Queues)
+	queues := normalizeQueues(params.queues)
 	orderedQueues := []string(nil)
-	if info.StrictPriority {
-		orderedQueues = sortByPriority(qcfg)
+	if params.strictPriority {
+		orderedQueues = sortByPriority(queues)
 	}
 	return &processor{
 		logger:         params.logger,
 		broker:         params.broker,
-		ss:             params.ss,
-		queueConfig:    qcfg,
+		queueConfig:    queues,
 		orderedQueues:  orderedQueues,
 		retryDelayFunc: params.retryDelayFunc,
 		syncRequestCh:  params.syncCh,
 		cancelations:   params.cancelations,
 		errLogLimiter:  rate.NewLimiter(rate.Every(3*time.Second), 1),
-		sema:           make(chan struct{}, info.Concurrency),
+		sema:           make(chan struct{}, params.concurrency),
 		done:           make(chan struct{}),
 		abort:          make(chan struct{}),
 		quit:           make(chan struct{}),
 		errHandler:     params.errHandler,
 		handler:        HandlerFunc(func(ctx context.Context, t *Task) error { return fmt.Errorf("handler not set") }),
+		starting:       params.starting,
+		finished:       params.finished,
 	}
 }
 
@@ -132,13 +137,9 @@ func (p *processor) terminate() {
 		p.sema <- struct{}{}
 	}
 	p.logger.Info("All workers have finished")
-	p.restore() // move any unfinished tasks back to the queue.
 }
 
 func (p *processor) start(wg *sync.WaitGroup) {
-	// NOTE: The call to "restore" needs to complete before starting
-	// the processor goroutine.
-	p.restore()
 	wg.Add(1)
 	go func() {
 		defer wg.Done()
@@ -161,14 +162,12 @@ func (p *processor) exec() {
 	msg, err := p.broker.Dequeue(qnames...)
 	switch {
 	case err == rdb.ErrNoProcessableTask:
-		// queues are empty, this is a normal behavior.
-		if len(qnames) > 1 {
-			// sleep to avoid slamming redis and let scheduler move tasks into queues.
-			// Note: With multiple queues, we are not using blocking pop operation and
-			// polling queues instead. This adds significant load to redis.
-			time.Sleep(time.Second)
-		}
 		p.logger.Debug("All queues are empty")
+		// Queues are empty, this is a normal behavior.
+		// Sleep to avoid slamming redis and let scheduler move tasks into queues.
+		// Note: We are not using blocking pop operation and polling queues instead.
+		// This adds significant load to redis.
+		time.Sleep(time.Second)
 		return
 	case err != nil:
 		if p.errLogLimiter.Allow() {
@@ -183,10 +182,10 @@ func (p *processor) exec() {
 		p.requeue(msg)
 		return
 	case p.sema <- struct{}{}: // acquire token
-		p.ss.AddWorkerStats(msg, time.Now())
+		p.starting <- msg
 		go func() {
 			defer func() {
-				p.ss.DeleteWorkerStats(msg)
+				p.finished <- msg
 				<-p.sema // release token
 			}()
 
@@ -203,8 +202,9 @@ func (p *processor) exec() {
 
 			select {
 			case <-p.quit:
-				// time is up, quit this worker goroutine.
+				// time is up, push the message back to queue and quit this worker goroutine.
 				p.logger.Warnf("Quitting worker. task id=%s", msg.ID)
+				p.requeue(msg)
 				return
 			case resErr := <-resCh:
 				// Note: One of three things should happen.
@@ -228,29 +228,19 @@ func (p *processor) exec() {
 	}
 }
 
-// restore moves all tasks from "in-progress" back to queue
-// to restore all unfinished tasks.
-func (p *processor) restore() {
-	n, err := p.broker.RequeueAll()
-	if err != nil {
-		p.logger.Errorf("Could not restore unfinished tasks: %v", err)
-	}
-	if n > 0 {
-		p.logger.Infof("Restored %d unfinished tasks back to queue", n)
-	}
-}
-
 func (p *processor) requeue(msg *base.TaskMessage) {
 	err := p.broker.Requeue(msg)
 	if err != nil {
 		p.logger.Errorf("Could not push task id=%s back to queue: %v", msg.ID, err)
+	} else {
+		p.logger.Infof("Pushed task id=%s back to queue", msg.ID)
 	}
 }
 
 func (p *processor) markAsDone(msg *base.TaskMessage) {
 	err := p.broker.Done(msg)
 	if err != nil {
-		errMsg := fmt.Sprintf("Could not remove task id=%s from %q", msg.ID, base.InProgressQueue)
+		errMsg := fmt.Sprintf("Could not remove task id=%s type=%q from %q err: %+v", msg.ID, msg.Type, base.InProgressQueue, err)
 		p.logger.Warnf("%s; Will retry syncing", errMsg)
 		p.syncRequestCh <- &syncRequest{
 			fn: func() error {
@@ -374,16 +364,15 @@ func (x byPriority) Len() int           { return len(x) }
 func (x byPriority) Less(i, j int) bool { return x[i].priority < x[j].priority }
 func (x byPriority) Swap(i, j int)      { x[i], x[j] = x[j], x[i] }
 
-// normalizeQueueCfg divides priority numbers by their
-// greatest common divisor.
-func normalizeQueueCfg(queueCfg map[string]int) map[string]int {
+// normalizeQueues divides priority numbers by their greatest common divisor.
+func normalizeQueues(queues map[string]int) map[string]int {
 	var xs []int
-	for _, x := range queueCfg {
+	for _, x := range queues {
 		xs = append(xs, x)
 	}
 	d := gcd(xs...)
 	res := make(map[string]int)
-	for q, x := range queueCfg {
+	for q, x := range queues {
 		res[q] = x / d
 	}
 	return res

processor_test.go

@@ -19,6 +19,29 @@ import (
 	"github.com/hibiken/asynq/internal/rdb"
 )
 
+// fakeHeartbeater receives from starting and finished channels and do nothing.
+func fakeHeartbeater(starting, finished <-chan *base.TaskMessage, done <-chan struct{}) {
+	for {
+		select {
+		case <-starting:
+		case <-finished:
+		case <-done:
+			return
+		}
+	}
+}
+
+// fakeSyncer receives from sync channel and do nothing.
+func fakeSyncer(syncCh <-chan *syncRequest, done <-chan struct{}) {
+	for {
+		select {
+		case <-syncCh:
+		case <-done:
+			return
+		}
+	}
+}
+
 func TestProcessorSuccess(t *testing.T) {
 	r := setup(t)
 	rdbClient := rdb.NewRDB(r)
@@ -63,16 +86,26 @@ func TestProcessorSuccess(t *testing.T) {
 			processed = append(processed, task)
 			return nil
 		}
-		ss := base.NewServerState("localhost", 1234, 10, defaultQueueConfig, false)
+		starting := make(chan *base.TaskMessage)
+		finished := make(chan *base.TaskMessage)
+		syncCh := make(chan *syncRequest)
+		done := make(chan struct{})
+		defer func() { close(done) }()
+		go fakeHeartbeater(starting, finished, done)
+		go fakeSyncer(syncCh, done)
 		p := newProcessor(processorParams{
 			logger:          testLogger,
 			broker:          rdbClient,
-			ss:              ss,
 			retryDelayFunc:  defaultDelayFunc,
-			syncCh:          nil,
+			syncCh:          syncCh,
 			cancelations:    base.NewCancelations(),
+			concurrency:     10,
+			queues:          defaultQueueConfig,
+			strictPriority:  false,
 			errHandler:      nil,
 			shutdownTimeout: defaultShutdownTimeout,
+			starting:        starting,
+			finished:        finished,
 		})
 		p.handler = HandlerFunc(handler)
 
@@ -84,16 +117,90 @@ func TestProcessorSuccess(t *testing.T) {
 				t.Fatal(err)
 			}
 		}
-		time.Sleep(time.Second) // wait for one second to allow all enqueued tasks to be processed.
-		p.terminate()
-
-		if diff := cmp.Diff(tc.wantProcessed, processed, sortTaskOpt, cmp.AllowUnexported(Payload{})); diff != "" {
-			t.Errorf("mismatch found in processed tasks; (-want, +got)\n%s", diff)
-		}
-
+		time.Sleep(2 * time.Second) // wait for two second to allow all enqueued tasks to be processed.
 		if l := r.LLen(base.InProgressQueue).Val(); l != 0 {
 			t.Errorf("%q has %d tasks, want 0", base.InProgressQueue, l)
 		}
+		p.terminate()
+
+		mu.Lock()
+		if diff := cmp.Diff(tc.wantProcessed, processed, sortTaskOpt, cmp.AllowUnexported(Payload{})); diff != "" {
+			t.Errorf("mismatch found in processed tasks; (-want, +got)\n%s", diff)
+		}
+		mu.Unlock()
+	}
+}
+
+// https://github.com/hibiken/asynq/issues/166
+func TestProcessTasksWithLargeNumberInPayload(t *testing.T) {
+	r := setup(t)
+	rdbClient := rdb.NewRDB(r)
+
+	m1 := h.NewTaskMessage("large_number", map[string]interface{}{"data": 111111111111111111})
+	t1 := NewTask(m1.Type, m1.Payload)
+
+	tests := []struct {
+		enqueued      []*base.TaskMessage // initial default queue state
+		wantProcessed []*Task             // tasks to be processed at the end
+	}{
+		{
+			enqueued:      []*base.TaskMessage{m1},
+			wantProcessed: []*Task{t1},
+		},
+	}
+
+	for _, tc := range tests {
+		h.FlushDB(t, r)                        // clean up db before each test case.
+		h.SeedEnqueuedQueue(t, r, tc.enqueued) // initialize default queue.
+
+		var mu sync.Mutex
+		var processed []*Task
+		handler := func(ctx context.Context, task *Task) error {
+			mu.Lock()
+			defer mu.Unlock()
+			if data, err := task.Payload.GetInt("data"); err != nil {
+				t.Errorf("coult not get data from payload: %v", err)
+			} else {
+				t.Logf("data == %d", data)
+			}
+			processed = append(processed, task)
+			return nil
+		}
+		starting := make(chan *base.TaskMessage)
+		finished := make(chan *base.TaskMessage)
+		syncCh := make(chan *syncRequest)
+		done := make(chan struct{})
+		defer func() { close(done) }()
+		go fakeHeartbeater(starting, finished, done)
+		go fakeSyncer(syncCh, done)
+		p := newProcessor(processorParams{
+			logger:          testLogger,
+			broker:          rdbClient,
+			retryDelayFunc:  defaultDelayFunc,
+			syncCh:          syncCh,
+			cancelations:    base.NewCancelations(),
+			concurrency:     10,
+			queues:          defaultQueueConfig,
+			strictPriority:  false,
+			errHandler:      nil,
+			shutdownTimeout: defaultShutdownTimeout,
+			starting:        starting,
+			finished:        finished,
+		})
+		p.handler = HandlerFunc(handler)
+
+		p.start(&sync.WaitGroup{})
+		time.Sleep(2 * time.Second) // wait for two second to allow all enqueued tasks to be processed.
+		if l := r.LLen(base.InProgressQueue).Val(); l != 0 {
+			t.Errorf("%q has %d tasks, want 0", base.InProgressQueue, l)
+		}
+		p.terminate()
+
+		mu.Lock()
+		if diff := cmp.Diff(tc.wantProcessed, processed, sortTaskOpt, cmpopts.IgnoreUnexported(Payload{})); diff != "" {
+			t.Errorf("mismatch found in processed tasks; (-want, +got)\n%s", diff)
+		}
+		mu.Unlock()
 	}
 }
 
@@ -140,7 +247,7 @@ func TestProcessorRetry(t *testing.T) {
 			handler: HandlerFunc(func(ctx context.Context, task *Task) error {
 				return fmt.Errorf(errMsg)
 			}),
-			wait: time.Second,
+			wait: 2 * time.Second,
 			wantRetry: []h.ZSetEntry{
 				{Msg: &r2, Score: float64(now.Add(time.Minute).Unix())},
 				{Msg: &r3, Score: float64(now.Add(time.Minute).Unix())},
@@ -168,16 +275,24 @@ func TestProcessorRetry(t *testing.T) {
 			defer mu.Unlock()
 			n++
 		}
-		ss := base.NewServerState("localhost", 1234, 10, defaultQueueConfig, false)
+		starting := make(chan *base.TaskMessage)
+		finished := make(chan *base.TaskMessage)
+		done := make(chan struct{})
+		defer func() { close(done) }()
+		go fakeHeartbeater(starting, finished, done)
 		p := newProcessor(processorParams{
 			logger:          testLogger,
 			broker:          rdbClient,
-			ss:              ss,
 			retryDelayFunc:  delayFunc,
 			syncCh:          nil,
 			cancelations:    base.NewCancelations(),
+			concurrency:     10,
+			queues:          defaultQueueConfig,
+			strictPriority:  false,
 			errHandler:      ErrorHandlerFunc(errHandler),
 			shutdownTimeout: defaultShutdownTimeout,
+			starting:        starting,
+			finished:        finished,
 		})
 		p.handler = tc.handler
 
@@ -189,7 +304,7 @@ func TestProcessorRetry(t *testing.T) {
 				t.Fatal(err)
 			}
 		}
-		time.Sleep(tc.wait)
+		time.Sleep(tc.wait) // FIXME: This makes test flaky.
 		p.terminate()
 
 		cmpOpt := cmpopts.EquateApprox(0, float64(time.Second)) // allow up to a second difference in zset score
@@ -241,16 +356,24 @@ func TestProcessorQueues(t *testing.T) {
 	}
 
 	for _, tc := range tests {
-		ss := base.NewServerState("localhost", 1234, 10, tc.queueCfg, false)
+		starting := make(chan *base.TaskMessage)
+		finished := make(chan *base.TaskMessage)
+		done := make(chan struct{})
+		defer func() { close(done) }()
+		go fakeHeartbeater(starting, finished, done)
 		p := newProcessor(processorParams{
 			logger:          testLogger,
 			broker:          nil,
-			ss:              ss,
 			retryDelayFunc:  defaultDelayFunc,
 			syncCh:          nil,
 			cancelations:    base.NewCancelations(),
+			concurrency:     10,
+			queues:          tc.queueCfg,
+			strictPriority:  false,
 			errHandler:      nil,
 			shutdownTimeout: defaultShutdownTimeout,
+			starting:        starting,
+			finished:        finished,
 		})
 		got := p.queues()
 		if diff := cmp.Diff(tc.want, got, sortOpt); diff != "" {
@@ -316,17 +439,24 @@ func TestProcessorWithStrictPriority(t *testing.T) {
 			base.DefaultQueueName: 2,
 			"low":                 1,
 		}
-		// Note: Set concurrency to 1 to make sure tasks are processed one at a time.
-		ss := base.NewServerState("localhost", 1234, 1 /* concurrency */, queueCfg, true /*strict*/)
+		starting := make(chan *base.TaskMessage)
+		finished := make(chan *base.TaskMessage)
+		done := make(chan struct{})
+		defer func() { close(done) }()
+		go fakeHeartbeater(starting, finished, done)
 		p := newProcessor(processorParams{
 			logger:          testLogger,
 			broker:          rdbClient,
-			ss:              ss,
 			retryDelayFunc:  defaultDelayFunc,
 			syncCh:          nil,
 			cancelations:    base.NewCancelations(),
+			concurrency:     1, // Set concurrency to 1 to make sure tasks are processed one at a time.
+			queues:          queueCfg,
+			strictPriority:  true,
 			errHandler:      nil,
 			shutdownTimeout: defaultShutdownTimeout,
+			starting:        starting,
+			finished:        finished,
 		})
 		p.handler = HandlerFunc(handler)
 
@@ -412,7 +542,7 @@ func TestGCD(t *testing.T) {
 	}
 }
 
-func TestNormalizeQueueCfg(t *testing.T) {
+func TestNormalizeQueues(t *testing.T) {
 	tests := []struct {
 		input map[string]int
 		want  map[string]int
@@ -462,9 +592,9 @@ func TestNormalizeQueueCfg(t *testing.T) {
 	}
 
 	for _, tc := range tests {
-		got := normalizeQueueCfg(tc.input)
+		got := normalizeQueues(tc.input)
 		if diff := cmp.Diff(tc.want, got); diff != "" {
-			t.Errorf("normalizeQueueCfg(%v) = %v, want %v; (-want, +got):\n%s",
+			t.Errorf("normalizeQueues(%v) = %v, want %v; (-want, +got):\n%s",
 				tc.input, got, tc.want, diff)
 		}
 	}

scheduler.go

@@ -21,29 +21,20 @@ type scheduler struct {
 
 	// poll interval on average
 	avgInterval time.Duration
-
-	// list of queues to move the tasks into.
-	qnames []string
 }
 
 type schedulerParams struct {
 	logger   *log.Logger
 	broker   base.Broker
 	interval time.Duration
-	queues   map[string]int
 }
 
 func newScheduler(params schedulerParams) *scheduler {
-	var qnames []string
-	for q := range params.queues {
-		qnames = append(qnames, q)
-	}
 	return &scheduler{
 		logger:      params.logger,
 		broker:      params.broker,
 		done:        make(chan struct{}),
 		avgInterval: params.interval,
-		qnames:      qnames,
 	}
 }
 
@@ -71,7 +62,7 @@ func (s *scheduler) start(wg *sync.WaitGroup) {
 }
 
 func (s *scheduler) exec() {
-	if err := s.broker.CheckAndEnqueue(s.qnames...); err != nil {
+	if err := s.broker.CheckAndEnqueue(); err != nil {
 		s.logger.Errorf("Could not enqueue scheduled tasks: %v", err)
 	}
 }

scheduler_test.go

@@ -23,7 +23,6 @@ func TestScheduler(t *testing.T) {
 		logger:   testLogger,
 		broker:   rdbClient,
 		interval: pollInterval,
-		queues:   defaultQueueConfig,
 	})
 	t1 := h.NewTaskMessage("gen_thumbnail", nil)
 	t2 := h.NewTaskMessage("send_email", nil)

server.go

@@ -10,7 +10,6 @@ import (
 	"fmt"
 	"math"
 	"math/rand"
-	"os"
 	"runtime"
 	"strings"
 	"sync"
@@ -34,12 +33,12 @@ import (
 // (e.g., queue size reaches a certain limit, or the task has been in the
 // queue for a certain amount of time).
 type Server struct {
-	ss *base.ServerState
-
 	logger *log.Logger
 
 	broker base.Broker
 
+	status *base.ServerStatus
+
 	// wait group to wait for all goroutines to finish.
 	wg          sync.WaitGroup
 	scheduler   *scheduler
@@ -283,15 +282,11 @@ func NewServer(r RedisConnOpt, cfg Config) *Server {
 	}
 	logger.SetLevel(toInternalLogLevel(loglevel))
 
-	host, err := os.Hostname()
-	if err != nil {
-		host = "unknown-host"
-	}
-	pid := os.Getpid()
-
 	rdb := rdb.NewRDB(createRedisClient(r))
-	ss := base.NewServerState(host, pid, n, queues, cfg.StrictPriority)
+	starting := make(chan *base.TaskMessage)
+	finished := make(chan *base.TaskMessage)
 	syncCh := make(chan *syncRequest)
+	status := base.NewServerStatus(base.StatusIdle)
 	cancels := base.NewCancelations()
 
 	syncer := newSyncer(syncerParams{
@@ -302,14 +297,18 @@ func NewServer(r RedisConnOpt, cfg Config) *Server {
 	heartbeater := newHeartbeater(heartbeaterParams{
 		logger:         logger,
 		broker:         rdb,
-		serverState: ss,
 		interval:       5 * time.Second,
+		concurrency:    n,
+		queues:         queues,
+		strictPriority: cfg.StrictPriority,
+		status:         status,
+		starting:       starting,
+		finished:       finished,
 	})
 	scheduler := newScheduler(schedulerParams{
 		logger:   logger,
 		broker:   rdb,
 		interval: 5 * time.Second,
-		queues:   queues,
 	})
 	subscriber := newSubscriber(subscriberParams{
 		logger:       logger,
@@ -319,17 +318,21 @@ func NewServer(r RedisConnOpt, cfg Config) *Server {
 	processor := newProcessor(processorParams{
 		logger:          logger,
 		broker:          rdb,
-		ss:              ss,
 		retryDelayFunc:  delayFunc,
 		syncCh:          syncCh,
 		cancelations:    cancels,
+		concurrency:     n,
+		queues:          queues,
+		strictPriority:  cfg.StrictPriority,
 		errHandler:      cfg.ErrorHandler,
 		shutdownTimeout: shutdownTimeout,
+		starting:        starting,
+		finished:        finished,
 	})
 	return &Server{
-		ss:          ss,
 		logger:      logger,
 		broker:      rdb,
+		status:      status,
 		scheduler:   scheduler,
 		processor:   processor,
 		syncer:      syncer,
@@ -390,13 +393,13 @@ func (srv *Server) Start(handler Handler) error {
 	if handler == nil {
 		return fmt.Errorf("asynq: server cannot run with nil handler")
 	}
-	switch srv.ss.Status() {
+	switch srv.status.Get() {
 	case base.StatusRunning:
 		return fmt.Errorf("asynq: the server is already running")
 	case base.StatusStopped:
 		return ErrServerStopped
 	}
-	srv.ss.SetStatus(base.StatusRunning)
+	srv.status.Set(base.StatusRunning)
 	srv.processor.handler = handler
 
 	srv.logger.Info("Starting processing")
@@ -414,7 +417,7 @@ func (srv *Server) Start(handler Handler) error {
 // active workers to finish processing tasks for duration specified in Config.ShutdownTimeout.
 // If worker didn't finish processing a task during the timeout, the task will be pushed back to Redis.
 func (srv *Server) Stop() {
-	switch srv.ss.Status() {
+	switch srv.status.Get() {
 	case base.StatusIdle, base.StatusStopped:
 		// server is not running, do nothing and return.
 		return
@@ -424,6 +427,7 @@ func (srv *Server) Stop() {
 	// Note: The order of termination is important.
 	// Sender goroutines should be terminated before the receiver goroutines.
 	// processor -> syncer (via syncCh)
+	// processor -> heartbeater (via starting, finished channels)
 	srv.scheduler.terminate()
 	srv.processor.terminate()
 	srv.syncer.terminate()
@@ -433,7 +437,7 @@ func (srv *Server) Stop() {
 	srv.wg.Wait()
 
 	srv.broker.Close()
-	srv.ss.SetStatus(base.StatusStopped)
+	srv.status.Set(base.StatusStopped)
 
 	srv.logger.Info("Exiting")
 }
@@ -443,6 +447,6 @@ func (srv *Server) Stop() {
 func (srv *Server) Quiet() {
 	srv.logger.Info("Stopping processor")
 	srv.processor.stop()
-	srv.ss.SetStatus(base.StatusQuiet)
+	srv.status.Set(base.StatusQuiet)
 	srv.logger.Info("Processor stopped")
 }

tools/asynq/README.md

@@ -14,6 +14,7 @@ Asynq CLI is a command line tool to monitor the tasks managed by `asynq` package
   - [Delete](#delete)
   - [Kill](#kill)
   - [Cancel](#cancel)
+  - [Pause](#pause)
 - [Config File](#config-file)
 
 ## Installation
@@ -144,6 +145,17 @@ Example:
 
     asynq cancel bnogo8gt6toe23vhef0g
 
+### Pause
+
+Command `pause` pauses the spcified queue. Tasks in paused queues are not processed by servers.
+To resume processing from the queue, use `unpause` command.
+To see which queues are currently paused, use `stats` command.
+
+Example:
+
+    asynq pause email
+    asynq unpause email
+
 ## Config File
 
 You can use a config file to set default values for the flags.

tools/asynq/cmd/pause.go

@@ -0,0 +1,47 @@
+// Copyright 2020 Kentaro Hibino. All rights reserved.
+// Use of this source code is governed by a MIT license
+// that can be found in the LICENSE file.
+
+package cmd
+
+import (
+	"fmt"
+	"os"
+
+	"github.com/go-redis/redis/v7"
+	"github.com/hibiken/asynq/internal/rdb"
+	"github.com/spf13/cobra"
+	"github.com/spf13/viper"
+)
+
+// pauseCmd represents the pause command
+var pauseCmd = &cobra.Command{
+	Use:   "pause [queue name]",
+	Short: "Pauses the specified queue",
+	Long: `Pause (asynq pause) will pause the specified queue.
+Asynq servers will not process tasks from paused queues.
+Use the "unpause" command to resume a paused queue.
+
+Example: asynq pause default -> Pause the "default" queue`,
+	Args: cobra.ExactValidArgs(1),
+	Run:  pause,
+}
+
+func init() {
+	rootCmd.AddCommand(pauseCmd)
+}
+
+func pause(cmd *cobra.Command, args []string) {
+	c := redis.NewClient(&redis.Options{
+		Addr:     viper.GetString("uri"),
+		DB:       viper.GetInt("db"),
+		Password: viper.GetString("password"),
+	})
+	r := rdb.NewRDB(c)
+	err := r.Pause(args[0])
+	if err != nil {
+		fmt.Printf("error: %v\n", err)
+		os.Exit(1)
+	}
+	fmt.Printf("Successfully paused queue %q\n", args[0])
+}

tools/asynq/cmd/stats.go

@@ -7,7 +7,6 @@ package cmd
 import (
 	"fmt"
 	"os"
-	"sort"
 	"strconv"
 	"strings"
 	"text/tabwriter"
@@ -96,24 +95,31 @@ func printStates(s *rdb.Stats) {
 	tw.Flush()
 }
 
-func printQueues(queues map[string]int) {
-	var qnames, seps, counts []string
-	for q := range queues {
-		qnames = append(qnames, strings.Title(q))
+func printQueues(queues []*rdb.Queue) {
+	var headers, seps, counts []string
+	for _, q := range queues {
+		title := queueTitle(q)
+		headers = append(headers, title)
+		seps = append(seps, strings.Repeat("-", len(title)))
+		counts = append(counts, strconv.Itoa(q.Size))
 	}
-	sort.Strings(qnames) // sort for stable order
-	for _, q := range qnames {
-		seps = append(seps, strings.Repeat("-", len(q)))
-		counts = append(counts, strconv.Itoa(queues[strings.ToLower(q)]))
-	}
-	format := strings.Repeat("%v\t", len(qnames)) + "\n"
+	format := strings.Repeat("%v\t", len(headers)) + "\n"
 	tw := new(tabwriter.Writer).Init(os.Stdout, 0, 8, 2, ' ', 0)
-	fmt.Fprintf(tw, format, toInterfaceSlice(qnames)...)
+	fmt.Fprintf(tw, format, toInterfaceSlice(headers)...)
 	fmt.Fprintf(tw, format, toInterfaceSlice(seps)...)
 	fmt.Fprintf(tw, format, toInterfaceSlice(counts)...)
 	tw.Flush()
 }
 
+func queueTitle(q *rdb.Queue) string {
+	var b strings.Builder
+	b.WriteString(strings.Title(q.Name))
+	if q.Paused {
+		b.WriteString(" (Paused)")
+	}
+	return b.String()
+}
+
 func printStats(s *rdb.Stats) {
 	format := strings.Repeat("%v\t", 3) + "\n"
 	tw := new(tabwriter.Writer).Init(os.Stdout, 0, 8, 2, ' ', 0)

tools/asynq/cmd/unpause.go

@@ -0,0 +1,46 @@
+// Copyright 2020 Kentaro Hibino. All rights reserved.
+// Use of this source code is governed by a MIT license
+// that can be found in the LICENSE file.
+
+package cmd
+
+import (
+	"fmt"
+	"os"
+
+	"github.com/go-redis/redis/v7"
+	"github.com/hibiken/asynq/internal/rdb"
+	"github.com/spf13/cobra"
+	"github.com/spf13/viper"
+)
+
+// unpauseCmd represents the unpause command
+var unpauseCmd = &cobra.Command{
+	Use:   "unpause [queue name]",
+	Short: "Unpauses the specified queue",
+	Long: `Unpause (asynq unpause) will unpause the specified queue.
+Asynq servers will process tasks from unpaused/resumed queues.
+
+Example: asynq unpause default -> Resume the "default" queue`,
+	Args: cobra.ExactValidArgs(1),
+	Run:  unpause,
+}
+
+func init() {
+	rootCmd.AddCommand(unpauseCmd)
+}
+
+func unpause(cmd *cobra.Command, args []string) {
+	c := redis.NewClient(&redis.Options{
+		Addr:     viper.GetString("uri"),
+		DB:       viper.GetInt("db"),
+		Password: viper.GetString("password"),
+	})
+	r := rdb.NewRDB(c)
+	err := r.Unpause(args[0])
+	if err != nil {
+		fmt.Printf("error: %v\n", err)
+		os.Exit(1)
+	}
+	fmt.Printf("Successfully resumed queue %q\n", args[0])
+}

tools/asynq/cmd/workers.go

@@ -61,7 +61,7 @@ func workers(cmd *cobra.Command, args []string) {
 		if x.Started != y.Started {
 			return x.Started.Before(y.Started)
 		}
-		return x.ID.String() < y.ID.String()
+		return x.ID < y.ID
 	})
 
 	cols := []string{"Process", "ID", "Type", "Payload", "Queue", "Started"}