Merge branch 'master' into v9

Signed-off-by: monkey <golang@88.com>

.github/dependabot.yml

@@ -0,0 +1,10 @@
+version: 2
+updates:
+- package-ecosystem: gomod
+  directory: /
+  schedule:
+    interval: weekly
+- package-ecosystem: github-actions
+  directory: /
+  schedule:
+    interval: weekly

README.md

@@ -166,3 +166,11 @@ Lastly, run:
 ```
 go test
 ```
+
+## Contributors
+
+Thanks to all the people who already contributed!
+
+<a href="https://github.com/go-redis/redis/graphs/contributors">
+  <img src="https://contributors-img.web.app/image?repo=go-redis/redis" />
+</a>

cluster.go

@@ -68,6 +68,9 @@ type ClusterOptions struct {
 	ReadTimeout  time.Duration
 	WriteTimeout time.Duration
 
+	// PoolFIFO uses FIFO mode for each node connection pool GET/PUT (default LIFO).
+	PoolFIFO bool
+
 	// PoolSize applies per cluster node and not for the whole cluster.
 	PoolSize           int
 	MinIdleConns       int
@@ -146,6 +149,7 @@ func (opt *ClusterOptions) clientOptions() *Options {
 		ReadTimeout:  opt.ReadTimeout,
 		WriteTimeout: opt.WriteTimeout,
 
+		PoolFIFO:           opt.PoolFIFO,
 		PoolSize:           opt.PoolSize,
 		MinIdleConns:       opt.MinIdleConns,
 		MaxConnAge:         opt.MaxConnAge,
@@ -591,8 +595,16 @@ func (c *clusterState) slotRandomNode(slot int) (*clusterNode, error) {
 	if len(nodes) == 0 {
 		return c.nodes.Random()
 	}
-	n := rand.Intn(len(nodes))
-	return nodes[n], nil
+	if len(nodes) == 1 {
+		return nodes[0], nil
+	}
+	randomNodes := rand.Perm(len(nodes))
+	for _, idx := range randomNodes {
+		if node := nodes[idx]; !node.Failing() {
+			return node, nil
+		}
+	}
+	return nodes[randomNodes[0]], nil
 }
 
 func (c *clusterState) slotNodes(slot int) []*clusterNode {

cluster_commands.go

@@ -8,55 +8,63 @@ import (
 
 func (c *ClusterClient) DBSize(ctx context.Context) *IntCmd {
 	cmd := NewIntCmd(ctx, "dbsize")
-	var size int64
-	err := c.ForEachMaster(ctx, func(ctx context.Context, master *Client) error {
-		n, err := master.DBSize(ctx).Result()
+	_ = c.hooks.process(ctx, cmd, func(ctx context.Context, _ Cmder) error {
+		var size int64
+		err := c.ForEachMaster(ctx, func(ctx context.Context, master *Client) error {
+			n, err := master.DBSize(ctx).Result()
+			if err != nil {
+				return err
+			}
+			atomic.AddInt64(&size, n)
+			return nil
+		})
 		if err != nil {
-			return err
+			cmd.SetErr(err)
+		} else {
+			cmd.val = size
 		}
-		atomic.AddInt64(&size, n)
 		return nil
 	})
-	if err != nil {
-		cmd.SetErr(err)
-		return cmd
-	}
-	cmd.val = size
 	return cmd
 }
 
 func (c *ClusterClient) ScriptLoad(ctx context.Context, script string) *StringCmd {
 	cmd := NewStringCmd(ctx, "script", "load", script)
-	mu := &sync.Mutex{}
-	err := c.ForEachShard(ctx, func(ctx context.Context, shard *Client) error {
-		val, err := shard.ScriptLoad(ctx, script).Result()
+	_ = c.hooks.process(ctx, cmd, func(ctx context.Context, _ Cmder) error {
+		mu := &sync.Mutex{}
+		err := c.ForEachShard(ctx, func(ctx context.Context, shard *Client) error {
+			val, err := shard.ScriptLoad(ctx, script).Result()
+			if err != nil {
+				return err
+			}
+
+			mu.Lock()
+			if cmd.Val() == "" {
+				cmd.val = val
+			}
+			mu.Unlock()
+
+			return nil
+		})
 		if err != nil {
-			return err
+			cmd.SetErr(err)
 		}
-
-		mu.Lock()
-		if cmd.Val() == "" {
-			cmd.val = val
-		}
-		mu.Unlock()
-
 		return nil
 	})
-	if err != nil {
-		cmd.SetErr(err)
-	}
-
 	return cmd
 }
 
 func (c *ClusterClient) ScriptFlush(ctx context.Context) *StatusCmd {
 	cmd := NewStatusCmd(ctx, "script", "flush")
-	_ = c.ForEachShard(ctx, func(ctx context.Context, shard *Client) error {
-		shard.ScriptFlush(ctx)
-
+	_ = c.hooks.process(ctx, cmd, func(ctx context.Context, _ Cmder) error {
+		err := c.ForEachShard(ctx, func(ctx context.Context, shard *Client) error {
+			return shard.ScriptFlush(ctx).Err()
+		})
+		if err != nil {
+			cmd.SetErr(err)
+		}
 		return nil
 	})
-
 	return cmd
 }
 
@@ -74,26 +82,28 @@ func (c *ClusterClient) ScriptExists(ctx context.Context, hashes ...string) *Boo
 		result[i] = true
 	}
 
-	mu := &sync.Mutex{}
-	err := c.ForEachShard(ctx, func(ctx context.Context, shard *Client) error {
-		val, err := shard.ScriptExists(ctx, hashes...).Result()
+	_ = c.hooks.process(ctx, cmd, func(ctx context.Context, _ Cmder) error {
+		mu := &sync.Mutex{}
+		err := c.ForEachShard(ctx, func(ctx context.Context, shard *Client) error {
+			val, err := shard.ScriptExists(ctx, hashes...).Result()
+			if err != nil {
+				return err
+			}
+
+			mu.Lock()
+			for i, v := range val {
+				result[i] = result[i] && v
+			}
+			mu.Unlock()
+
+			return nil
+		})
 		if err != nil {
-			return err
+			cmd.SetErr(err)
+		} else {
+			cmd.val = result
 		}
-
-		mu.Lock()
-		for i, v := range val {
-			result[i] = result[i] && v
-		}
-		mu.Unlock()
-
 		return nil
 	})
-	if err != nil {
-		cmd.SetErr(err)
-	}
-
-	cmd.val = result
-
 	return cmd
 }

commands.go

@@ -190,6 +190,7 @@ type Cmdable interface {
 	LSet(ctx context.Context, key string, index int64, value interface{}) *StatusCmd
 	LTrim(ctx context.Context, key string, start, stop int64) *StatusCmd
 	RPop(ctx context.Context, key string) *StringCmd
+	RPopCount(ctx context.Context, key string, count int) *StringSliceCmd
 	RPopLPush(ctx context.Context, source, destination string) *StringCmd
 	RPush(ctx context.Context, key string, values ...interface{}) *IntCmd
 	RPushX(ctx context.Context, key string, values ...interface{}) *IntCmd
@@ -1452,6 +1453,12 @@ func (c cmdable) RPop(ctx context.Context, key string) *StringCmd {
 	return cmd
 }
 
+func (c cmdable) RPopCount(ctx context.Context, key string, count int) *StringSliceCmd {
+	cmd := NewStringSliceCmd(ctx, "rpop", key, count)
+	_ = c(ctx, cmd)
+	return cmd
+}
+
 func (c cmdable) RPopLPush(ctx context.Context, source, destination string) *StringCmd {
 	cmd := NewStringCmd(ctx, "rpoplpush", source, destination)
 	_ = c(ctx, cmd)

commands_test.go

@@ -2282,6 +2282,20 @@ var _ = Describe("Commands", func() {
 			Expect(lRange.Val()).To(Equal([]string{"one", "two"}))
 		})
 
+		It("should RPopCount", func() {
+			rPush := client.RPush(ctx, "list", "one", "two", "three", "four")
+			Expect(rPush.Err()).NotTo(HaveOccurred())
+			Expect(rPush.Val()).To(Equal(int64(4)))
+
+			rPopCount := client.RPopCount(ctx, "list", 2)
+			Expect(rPopCount.Err()).NotTo(HaveOccurred())
+			Expect(rPopCount.Val()).To(Equal([]string{"four", "three"}))
+
+			lRange := client.LRange(ctx, "list", 0, -1)
+			Expect(lRange.Err()).NotTo(HaveOccurred())
+			Expect(lRange.Val()).To(Equal([]string{"one", "two"}))
+		})
+
 		It("should RPopLPush", func() {
 			rPush := client.RPush(ctx, "list", "one")
 			Expect(rPush.Err()).NotTo(HaveOccurred())
@@ -4113,6 +4127,45 @@ var _ = Describe("Commands", func() {
 			}))
 		})
 
+		It("should ZUnion", func() {
+			err := client.ZAddArgs(ctx, "zset1", redis.ZAddArgs{
+				Members: []redis.Z{
+					{Score: 1, Member: "one"},
+					{Score: 2, Member: "two"},
+				},
+			}).Err()
+			Expect(err).NotTo(HaveOccurred())
+
+			err = client.ZAddArgs(ctx, "zset2", redis.ZAddArgs{
+				Members: []redis.Z{
+					{Score: 1, Member: "one"},
+					{Score: 2, Member: "two"},
+					{Score: 3, Member: "three"},
+				},
+			}).Err()
+			Expect(err).NotTo(HaveOccurred())
+
+			union, err := client.ZUnion(ctx, redis.ZStore{
+				Keys:      []string{"zset1", "zset2"},
+				Weights:   []float64{2, 3},
+				Aggregate: "sum",
+			}).Result()
+			Expect(err).NotTo(HaveOccurred())
+			Expect(union).To(Equal([]string{"one", "three", "two"}))
+
+			unionScores, err := client.ZUnionWithScores(ctx, redis.ZStore{
+				Keys:      []string{"zset1", "zset2"},
+				Weights:   []float64{2, 3},
+				Aggregate: "sum",
+			}).Result()
+			Expect(err).NotTo(HaveOccurred())
+			Expect(unionScores).To(Equal([]redis.Z{
+				{Score: 5, Member: "one"},
+				{Score: 9, Member: "three"},
+				{Score: 10, Member: "two"},
+			}))
+		})
+
 		It("should ZUnionStore", func() {
 			err := client.ZAdd(ctx, "zset1", redis.Z{Score: 1, Member: "one"}).Err()
 			Expect(err).NotTo(HaveOccurred())
@@ -4339,6 +4392,33 @@ var _ = Describe("Commands", func() {
 			Expect(n).To(Equal(int64(3)))
 		})
 
+		// TODO XTrimMaxLenApprox/XTrimMinIDApprox There is a bug in the limit parameter.
+		// TODO Don't test it for now.
+		// TODO link: https://github.com/redis/redis/issues/9046
+		It("should XTrimMaxLen", func() {
+			n, err := client.XTrimMaxLen(ctx, "stream", 0).Result()
+			Expect(err).NotTo(HaveOccurred())
+			Expect(n).To(Equal(int64(3)))
+		})
+
+		It("should XTrimMaxLenApprox", func() {
+			n, err := client.XTrimMaxLenApprox(ctx, "stream", 0, 0).Result()
+			Expect(err).NotTo(HaveOccurred())
+			Expect(n).To(Equal(int64(3)))
+		})
+
+		It("should XTrimMinID", func() {
+			n, err := client.XTrimMinID(ctx, "stream", "4-0").Result()
+			Expect(err).NotTo(HaveOccurred())
+			Expect(n).To(Equal(int64(3)))
+		})
+
+		It("should XTrimMinIDApprox", func() {
+			n, err := client.XTrimMinIDApprox(ctx, "stream", "4-0", 0).Result()
+			Expect(err).NotTo(HaveOccurred())
+			Expect(n).To(Equal(int64(3)))
+		})
+
 		It("should XAdd", func() {
 			id, err := client.XAdd(ctx, &redis.XAddArgs{
 				Stream: "stream",

example_test.go

@@ -417,7 +417,7 @@ func ExampleClient_Watch() {
 			// Actual opperation (local in optimistic lock).
 			n++
 
-			// Operation is commited only if the watched keys remain unchanged.
+			// Operation is committed only if the watched keys remain unchanged.
 			_, err = tx.TxPipelined(ctx, func(pipe redis.Pipeliner) error {
 				pipe.Set(ctx, key, n, 0)
 				return nil

internal/pool/conncheck.go

@@ -0,0 +1,45 @@
+// +build linux darwin dragonfly freebsd netbsd openbsd solaris illumos
+
+package pool
+
+import (
+	"errors"
+	"io"
+	"net"
+	"syscall"
+)
+
+var errUnexpectedRead = errors.New("unexpected read from socket")
+
+func connCheck(conn net.Conn) error {
+	sysConn, ok := conn.(syscall.Conn)
+	if !ok {
+		return nil
+	}
+	rawConn, err := sysConn.SyscallConn()
+	if err != nil {
+		return err
+	}
+
+	var sysErr error
+	err = rawConn.Read(func(fd uintptr) bool {
+		var buf [1]byte
+		n, err := syscall.Read(int(fd), buf[:])
+		switch {
+		case n == 0 && err == nil:
+			sysErr = io.EOF
+		case n > 0:
+			sysErr = errUnexpectedRead
+		case err == syscall.EAGAIN || err == syscall.EWOULDBLOCK:
+			sysErr = nil
+		default:
+			sysErr = err
+		}
+		return true
+	})
+	if err != nil {
+		return err
+	}
+
+	return sysErr
+}

internal/pool/conncheck_dummy.go

@@ -0,0 +1,9 @@
+// +build !linux,!darwin,!dragonfly,!freebsd,!netbsd,!openbsd,!solaris,!illumos
+
+package pool
+
+import "net"
+
+func connCheck(conn net.Conn) error {
+	return nil
+}

internal/pool/conncheck_test.go

@@ -0,0 +1,46 @@
+// +build linux darwin dragonfly freebsd netbsd openbsd solaris illumos
+
+package pool
+
+import (
+	"net"
+	"net/http/httptest"
+	"testing"
+	"time"
+)
+
+func Test_connCheck(t *testing.T) {
+	// tests with real conns
+	ts := httptest.NewServer(nil)
+	defer ts.Close()
+
+	t.Run("good conn", func(t *testing.T) {
+		conn, err := net.DialTimeout(ts.Listener.Addr().Network(), ts.Listener.Addr().String(), time.Second)
+		if err != nil {
+			t.Fatalf(err.Error())
+		}
+		defer conn.Close()
+		if err = connCheck(conn); err != nil {
+			t.Fatalf(err.Error())
+		}
+		conn.Close()
+
+		if err = connCheck(conn); err == nil {
+			t.Fatalf("expect has error")
+		}
+	})
+
+	t.Run("bad conn 2", func(t *testing.T) {
+		conn, err := net.DialTimeout(ts.Listener.Addr().Network(), ts.Listener.Addr().String(), time.Second)
+		if err != nil {
+			t.Fatalf(err.Error())
+		}
+		defer conn.Close()
+
+		ts.Close()
+
+		if err = connCheck(conn); err == nil {
+			t.Fatalf("expect has err")
+		}
+	})
+}

internal/pool/main_test.go

@@ -2,9 +2,12 @@ package pool_test
 
 import (
 	"context"
+	"fmt"
 	"net"
 	"sync"
+	"syscall"
 	"testing"
+	"time"
 
 	. "github.com/onsi/ginkgo"
 	. "github.com/onsi/gomega"
@@ -32,5 +35,87 @@ func perform(n int, cbs ...func(int)) {
 }
 
 func dummyDialer(context.Context) (net.Conn, error) {
-	return &net.TCPConn{}, nil
+	// return &net.TCPConn{}, nil
+	return newDummyConn(), nil
+}
+
+func newDummyConn() net.Conn {
+	return &dummyConn{
+		rawConn: &dummyRawConn{},
+	}
+}
+
+var _ net.Conn = (*dummyConn)(nil)
+var _ syscall.Conn = (*dummyConn)(nil)
+
+type dummyConn struct {
+	rawConn *dummyRawConn
+}
+
+func (d *dummyConn) SyscallConn() (syscall.RawConn, error) {
+	return d.rawConn, nil
+}
+
+var errDummy = fmt.Errorf("dummyConn err")
+
+func (d *dummyConn) Read(b []byte) (n int, err error) {
+	return 0, errDummy
+}
+
+func (d *dummyConn) Write(b []byte) (n int, err error) {
+	return 0, errDummy
+}
+
+func (d *dummyConn) Close() error {
+	d.rawConn.Close()
+	return nil
+}
+
+func (d *dummyConn) LocalAddr() net.Addr {
+	return &net.TCPAddr{}
+}
+
+func (d *dummyConn) RemoteAddr() net.Addr {
+	return &net.TCPAddr{}
+}
+
+func (d *dummyConn) SetDeadline(t time.Time) error {
+	return nil
+}
+
+func (d *dummyConn) SetReadDeadline(t time.Time) error {
+	return nil
+}
+
+func (d *dummyConn) SetWriteDeadline(t time.Time) error {
+	return nil
+}
+
+var _ syscall.RawConn = (*dummyRawConn)(nil)
+
+type dummyRawConn struct {
+	closed bool
+	mux    sync.Mutex
+}
+
+func (d *dummyRawConn) Control(f func(fd uintptr)) error {
+	return nil
+}
+
+func (d *dummyRawConn) Read(f func(fd uintptr) (done bool)) error {
+	d.mux.Lock()
+	defer d.mux.Unlock()
+	if d.closed {
+		return fmt.Errorf("dummyRawConn closed")
+	}
+	return nil
+}
+
+func (d *dummyRawConn) Write(f func(fd uintptr) (done bool)) error {
+	return nil
+}
+func (d *dummyRawConn) Close() {
+	d.mux.Lock()
+	d.closed = true
+	d.mux.Unlock()
 }

internal/pool/pool.go

@@ -57,6 +57,7 @@ type Options struct {
 	Dialer  func(context.Context) (net.Conn, error)
 	OnClose func(*Conn) error
 
+	PoolFIFO           bool
 	PoolSize           int
 	MinIdleConns       int
 	MaxConnAge         time.Duration
@@ -308,13 +309,21 @@ func (p *ConnPool) freeTurn() {
 }
 
 func (p *ConnPool) popIdle() *Conn {
-	if len(p.idleConns) == 0 {
+	n := len(p.idleConns)
+	if n == 0 {
 		return nil
 	}
 
-	idx := len(p.idleConns) - 1
-	cn := p.idleConns[idx]
-	p.idleConns = p.idleConns[:idx]
+	var cn *Conn
+	if p.opt.PoolFIFO {
+		cn = p.idleConns[0]
+		copy(p.idleConns, p.idleConns[1:])
+		p.idleConns = p.idleConns[:n-1]
+	} else {
+		idx := n - 1
+		cn = p.idleConns[idx]
+		p.idleConns = p.idleConns[:idx]
+	}
 	p.idleConnsLen--
 	p.checkMinIdleConns()
 	return cn
@@ -511,7 +520,7 @@ func (p *ConnPool) reapStaleConn() *Conn {
 
 func (p *ConnPool) isStaleConn(cn *Conn) bool {
 	if p.opt.IdleTimeout == 0 && p.opt.MaxConnAge == 0 {
-		return false
+		return connCheck(cn.netConn) != nil
 	}
 
 	now := time.Now()
@@ -522,5 +531,5 @@ func (p *ConnPool) isStaleConn(cn *Conn) bool {
 		return true
 	}
 
-	return false
+	return connCheck(cn.netConn) != nil
 }

internal/pool/pool_test.go

@@ -6,10 +6,10 @@ import (
 	"testing"
 	"time"
 
-	"github.com/go-redis/redis/v8/internal/pool"
-
 	. "github.com/onsi/ginkgo"
 	. "github.com/onsi/gomega"
+
+	"github.com/go-redis/redis/v8/internal/pool"
 )
 
 var _ = Describe("ConnPool", func() {
@@ -285,6 +285,8 @@ var _ = Describe("conns reaper", func() {
 					cn.SetUsedAt(time.Now().Add(-2 * idleTimeout))
 				case "aged":
 					cn.SetCreatedAt(time.Now().Add(-2 * maxAge))
+				case "conncheck":
+					cn.Close()
 				}
 				conns = append(conns, cn)
 				staleConns = append(staleConns, cn)
@@ -371,6 +373,7 @@ var _ = Describe("conns reaper", func() {
 
 	assert("idle")
 	assert("aged")
+	assert("conncheck")
 })
 
 var _ = Describe("race", func() {

main_test.go

@@ -12,10 +12,10 @@ import (
 	"testing"
 	"time"
 
-	"github.com/go-redis/redis/v8"
-
 	. "github.com/onsi/ginkgo"
 	. "github.com/onsi/gomega"
+
+	"github.com/go-redis/redis/v8"
 )
 
 const (
@@ -117,7 +117,7 @@ func TestGinkgoSuite(t *testing.T) {
 	RunSpecs(t, "go-redis")
 }
 
-//------------------------------------------------------------------------------
+// ------------------------------------------------------------------------------
 
 func redisOptions() *redis.Options {
 	return &redis.Options{
@@ -364,7 +364,7 @@ func startSentinel(port, masterName, masterPort string) (*redisProcess, error) {
 	return p, nil
 }
 
-//------------------------------------------------------------------------------
+// ------------------------------------------------------------------------------
 
 type badConnError string
 
@@ -409,7 +409,7 @@ func (cn *badConn) Write([]byte) (int, error) {
 	return 0, badConnError("bad connection")
 }
 
-//------------------------------------------------------------------------------
+// ------------------------------------------------------------------------------
 
 type hook struct {
 	beforeProcess func(ctx context.Context, cmd redis.Cmder) (context.Context, error)

options.go

@@ -76,6 +76,10 @@ type Options struct {
 	// Default is ReadTimeout.
 	WriteTimeout time.Duration
 
+	// Type of connection pool.
+	// true for FIFO pool, false for LIFO pool.
+	// Note that fifo has higher overhead compared to lifo.
+	PoolFIFO bool
 	// Maximum number of socket connections.
 	// Default is 10 connections per every available CPU as reported by runtime.GOMAXPROCS.
 	PoolSize int
@@ -291,6 +295,7 @@ func newConnPool(opt *Options) *pool.ConnPool {
 		Dialer: func(ctx context.Context) (net.Conn, error) {
 			return opt.Dialer(ctx, opt.Network, opt.Addr)
 		},
+		PoolFIFO:           opt.PoolFIFO,
 		PoolSize:           opt.PoolSize,
 		MinIdleConns:       opt.MinIdleConns,
 		MaxConnAge:         opt.MaxConnAge,

pool_test.go

@@ -87,8 +87,9 @@ var _ = Describe("pool", func() {
 		cn.SetNetConn(&badConn{})
 		client.Pool().Put(ctx, cn)
 
+		// connCheck will automatically remove damaged connections.
 		err = client.Ping(ctx).Err()
-		Expect(err).To(MatchError("bad connection"))
+		Expect(err).NotTo(HaveOccurred())
 
 		val, err := client.Ping(ctx).Result()
 		Expect(err).NotTo(HaveOccurred())

ring.go

@@ -78,6 +78,9 @@ type RingOptions struct {
 	ReadTimeout  time.Duration
 	WriteTimeout time.Duration
 
+	// PoolFIFO uses FIFO mode for each node connection pool GET/PUT (default LIFO).
+	PoolFIFO bool
+
 	PoolSize           int
 	MinIdleConns       int
 	MaxConnAge         time.Duration
@@ -138,6 +141,7 @@ func (opt *RingOptions) clientOptions() *Options {
 		ReadTimeout:  opt.ReadTimeout,
 		WriteTimeout: opt.WriteTimeout,
 
+		PoolFIFO:           opt.PoolFIFO,
 		PoolSize:           opt.PoolSize,
 		MinIdleConns:       opt.MinIdleConns,
 		MaxConnAge:         opt.MaxConnAge,

sentinel.go

@@ -57,6 +57,9 @@ type FailoverOptions struct {
 	ReadTimeout  time.Duration
 	WriteTimeout time.Duration
 
+	// PoolFIFO uses FIFO mode for each node connection pool GET/PUT (default LIFO).
+	PoolFIFO bool
+
 	PoolSize           int
 	MinIdleConns       int
 	MaxConnAge         time.Duration
@@ -86,6 +89,7 @@ func (opt *FailoverOptions) clientOptions() *Options {
 		ReadTimeout:  opt.ReadTimeout,
 		WriteTimeout: opt.WriteTimeout,
 
+		PoolFIFO:           opt.PoolFIFO,
 		PoolSize:           opt.PoolSize,
 		PoolTimeout:        opt.PoolTimeout,
 		IdleTimeout:        opt.IdleTimeout,
@@ -115,6 +119,7 @@ func (opt *FailoverOptions) sentinelOptions(addr string) *Options {
 		ReadTimeout:  opt.ReadTimeout,
 		WriteTimeout: opt.WriteTimeout,
 
+		PoolFIFO:           opt.PoolFIFO,
 		PoolSize:           opt.PoolSize,
 		PoolTimeout:        opt.PoolTimeout,
 		IdleTimeout:        opt.IdleTimeout,
@@ -146,6 +151,7 @@ func (opt *FailoverOptions) clusterOptions() *ClusterOptions {
 		ReadTimeout:  opt.ReadTimeout,
 		WriteTimeout: opt.WriteTimeout,
 
+		PoolFIFO:           opt.PoolFIFO,
 		PoolSize:           opt.PoolSize,
 		PoolTimeout:        opt.PoolTimeout,
 		IdleTimeout:        opt.IdleTimeout,

sentinel_test.go

@@ -192,7 +192,7 @@ var _ = Describe("NewFailoverClusterClient", func() {
 		err = master.Shutdown(ctx).Err()
 		Expect(err).NotTo(HaveOccurred())
 		Eventually(func() error {
-			return sentinelMaster.Ping(ctx).Err()
+			return master.Ping(ctx).Err()
 		}, "15s", "100ms").Should(HaveOccurred())
 
 		// Check that client picked up new master.

tx_test.go

@@ -123,7 +123,7 @@ var _ = Describe("Tx", func() {
 		Expect(num).To(Equal(int64(N)))
 	})
 
-	It("should recover from bad connection", func() {
+	It("should remove from bad connection", func() {
 		// Put bad connection in the pool.
 		cn, err := client.Pool().Get(context.Background())
 		Expect(err).NotTo(HaveOccurred())
@@ -134,17 +134,14 @@ var _ = Describe("Tx", func() {
 		do := func() error {
 			err := client.Watch(ctx, func(tx *redis.Tx) error {
 				_, err := tx.TxPipelined(ctx, func(pipe redis.Pipeliner) error {
-					pipe.Ping(ctx)
-					return nil
+					return pipe.Ping(ctx).Err()
 				})
 				return err
 			})
 			return err
 		}
 
-		err = do()
-		Expect(err).To(MatchError("bad connection"))
-
+		// connCheck will automatically remove damaged connections.
 		err = do()
 		Expect(err).NotTo(HaveOccurred())
 	})

universal.go

@@ -35,6 +35,9 @@ type UniversalOptions struct {
 	ReadTimeout  time.Duration
 	WriteTimeout time.Duration
 
+	// PoolFIFO uses FIFO mode for each node connection pool GET/PUT (default LIFO).
+	PoolFIFO bool
+
 	PoolSize           int
 	MinIdleConns       int
 	MaxConnAge         time.Duration
@@ -83,6 +86,7 @@ func (o *UniversalOptions) Cluster() *ClusterOptions {
 		DialTimeout:        o.DialTimeout,
 		ReadTimeout:        o.ReadTimeout,
 		WriteTimeout:       o.WriteTimeout,
+		PoolFIFO:           o.PoolFIFO,
 		PoolSize:           o.PoolSize,
 		MinIdleConns:       o.MinIdleConns,
 		MaxConnAge:         o.MaxConnAge,
@@ -120,6 +124,7 @@ func (o *UniversalOptions) Failover() *FailoverOptions {
 		ReadTimeout:  o.ReadTimeout,
 		WriteTimeout: o.WriteTimeout,
 
+		PoolFIFO:           o.PoolFIFO,
 		PoolSize:           o.PoolSize,
 		MinIdleConns:       o.MinIdleConns,
 		MaxConnAge:         o.MaxConnAge,
@@ -155,6 +160,7 @@ func (o *UniversalOptions) Simple() *Options {
 		ReadTimeout:  o.ReadTimeout,
 		WriteTimeout: o.WriteTimeout,
 
+		PoolFIFO:           o.PoolFIFO,
 		PoolSize:           o.PoolSize,
 		MinIdleConns:       o.MinIdleConns,
 		MaxConnAge:         o.MaxConnAge,