diff --git a/cluster.go b/cluster.go
index 2759d8f..f962684 100644
--- a/cluster.go
+++ b/cluster.go
@@ -1180,8 +1180,8 @@ func (c *ClusterClient) processPipeline(ctx context.Context, cmds []Cmder) error
 
 func (c *ClusterClient) _processPipeline(ctx context.Context, cmds []Cmder) error {
 	cmdsMap := newCmdsMap()
-	err := c.mapCmdsByNode(ctx, cmdsMap, cmds)
-	if err != nil {
+
+	if err := c.mapCmdsByNode(ctx, cmdsMap, cmds); err != nil {
 		setCmdsErr(cmds, err)
 		return err
 	}
@@ -1201,18 +1201,7 @@ func (c *ClusterClient) _processPipeline(ctx context.Context, cmds []Cmder) erro
 			wg.Add(1)
 			go func(node *clusterNode, cmds []Cmder) {
 				defer wg.Done()
-
-				err := c._processPipelineNode(ctx, node, cmds, failedCmds)
-				if err == nil {
-					return
-				}
-				if attempt < c.opt.MaxRedirects {
-					if err := c.mapCmdsByNode(ctx, failedCmds, cmds); err != nil {
-						setCmdsErr(cmds, err)
-					}
-				} else {
-					setCmdsErr(cmds, err)
-				}
+				c._processPipelineNode(ctx, node, cmds, failedCmds)
 			}(node, cmds)
 		}
 
@@ -1267,13 +1256,13 @@ func (c *ClusterClient) cmdsAreReadOnly(ctx context.Context, cmds []Cmder) bool
 
 func (c *ClusterClient) _processPipelineNode(
 	ctx context.Context, node *clusterNode, cmds []Cmder, failedCmds *cmdsMap,
-) error {
-	return node.Client.hooks.processPipeline(ctx, cmds, func(ctx context.Context, cmds []Cmder) error {
+) {
+	_ = node.Client.hooks.processPipeline(ctx, cmds, func(ctx context.Context, cmds []Cmder) error {
 		return node.Client.withConn(ctx, func(ctx context.Context, cn *pool.Conn) error {
-			err := cn.WithWriter(ctx, c.opt.WriteTimeout, func(wr *proto.Writer) error {
+			if err := cn.WithWriter(ctx, c.opt.WriteTimeout, func(wr *proto.Writer) error {
 				return writeCmds(wr, cmds)
-			})
-			if err != nil {
+			}); err != nil {
+				setCmdsErr(cmds, err)
 				return err
 			}
 
@@ -1291,7 +1280,7 @@ func (c *ClusterClient) pipelineReadCmds(
 	cmds []Cmder,
 	failedCmds *cmdsMap,
 ) error {
-	for _, cmd := range cmds {
+	for i, cmd := range cmds {
 		err := cmd.readReply(rd)
 		cmd.SetErr(err)
 
@@ -1303,15 +1292,24 @@ func (c *ClusterClient) pipelineReadCmds(
 			continue
 		}
 
-		if c.opt.ReadOnly && (isLoadingError(err) || !isRedisError(err)) {
+		if c.opt.ReadOnly {
 			node.MarkAsFailing()
+		}
+
+		if !isRedisError(err) {
+			if shouldRetry(err, true) {
+				_ = c.mapCmdsByNode(ctx, failedCmds, cmds)
+			}
+			setCmdsErr(cmds[i+1:], err)
 			return err
 		}
-		if isRedisError(err) {
-			continue
-		}
+	}
+
+	if err := cmds[0].Err(); err != nil && shouldRetry(err, true) {
+		_ = c.mapCmdsByNode(ctx, failedCmds, cmds)
 		return err
 	}
+
 	return nil
 }
 
@@ -1393,19 +1391,7 @@ func (c *ClusterClient) _processTxPipeline(ctx context.Context, cmds []Cmder) er
 				wg.Add(1)
 				go func(node *clusterNode, cmds []Cmder) {
 					defer wg.Done()
-
-					err := c._processTxPipelineNode(ctx, node, cmds, failedCmds)
-					if err == nil {
-						return
-					}
-
-					if attempt < c.opt.MaxRedirects {
-						if err := c.mapCmdsByNode(ctx, failedCmds, cmds); err != nil {
-							setCmdsErr(cmds, err)
-						}
-					} else {
-						setCmdsErr(cmds, err)
-					}
+					c._processTxPipelineNode(ctx, node, cmds, failedCmds)
 				}(node, cmds)
 			}
 
@@ -1431,34 +1417,39 @@ func (c *ClusterClient) mapCmdsBySlot(ctx context.Context, cmds []Cmder) map[int
 
 func (c *ClusterClient) _processTxPipelineNode(
 	ctx context.Context, node *clusterNode, cmds []Cmder, failedCmds *cmdsMap,
-) error {
-	return node.Client.hooks.processTxPipeline(ctx, cmds, func(ctx context.Context, cmds []Cmder) error {
-		return node.Client.withConn(ctx, func(ctx context.Context, cn *pool.Conn) error {
-			err := cn.WithWriter(ctx, c.opt.WriteTimeout, func(wr *proto.Writer) error {
-				return writeCmds(wr, cmds)
-			})
-			if err != nil {
-				return err
-			}
-
-			return cn.WithReader(ctx, c.opt.ReadTimeout, func(rd *proto.Reader) error {
-				statusCmd := cmds[0].(*StatusCmd)
-				// Trim multi and exec.
-				cmds = cmds[1 : len(cmds)-1]
-
-				err := c.txPipelineReadQueued(ctx, rd, statusCmd, cmds, failedCmds)
-				if err != nil {
-					moved, ask, addr := isMovedError(err)
-					if moved || ask {
-						return c.cmdsMoved(ctx, cmds, moved, ask, addr, failedCmds)
-					}
+) {
+	_ = node.Client.hooks.processTxPipeline(
+		ctx, cmds, func(ctx context.Context, cmds []Cmder) error {
+			return node.Client.withConn(ctx, func(ctx context.Context, cn *pool.Conn) error {
+				if err := cn.WithWriter(ctx, c.opt.WriteTimeout, func(wr *proto.Writer) error {
+					return writeCmds(wr, cmds)
+				}); err != nil {
+					setCmdsErr(cmds, err)
 					return err
 				}
 
-				return pipelineReadCmds(rd, cmds)
+				return cn.WithReader(ctx, c.opt.ReadTimeout, func(rd *proto.Reader) error {
+					statusCmd := cmds[0].(*StatusCmd)
+					// Trim multi and exec.
+					trimmedCmds := cmds[1 : len(cmds)-1]
+
+					if err := c.txPipelineReadQueued(
+						ctx, rd, statusCmd, trimmedCmds, failedCmds,
+					); err != nil {
+						setCmdsErr(cmds, err)
+
+						moved, ask, addr := isMovedError(err)
+						if moved || ask {
+							return c.cmdsMoved(ctx, trimmedCmds, moved, ask, addr, failedCmds)
+						}
+
+						return err
+					}
+
+					return pipelineReadCmds(rd, trimmedCmds)
+				})
 			})
 		})
-	})
 }
 
 func (c *ClusterClient) txPipelineReadQueued(
diff --git a/cluster_test.go b/cluster_test.go
index 72938a2..92844eb 100644
--- a/cluster_test.go
+++ b/cluster_test.go
@@ -1276,20 +1276,33 @@ var _ = Describe("ClusterClient timeout", func() {
 	Context("read/write timeout", func() {
 		BeforeEach(func() {
 			opt := redisClusterOptions()
-			opt.ReadTimeout = 250 * time.Millisecond
-			opt.WriteTimeout = 250 * time.Millisecond
-			opt.MaxRedirects = 1
 			client = cluster.newClusterClient(ctx, opt)
 
 			err := client.ForEachShard(ctx, func(ctx context.Context, client *redis.Client) error {
-				return client.ClientPause(ctx, pause).Err()
+				err := client.ClientPause(ctx, pause).Err()
+
+				opt := client.Options()
+				opt.ReadTimeout = time.Nanosecond
+				opt.WriteTimeout = time.Nanosecond
+
+				return err
 			})
 			Expect(err).NotTo(HaveOccurred())
+
+			// Overwrite timeouts after the client is initialized.
+			opt.ReadTimeout = time.Nanosecond
+			opt.WriteTimeout = time.Nanosecond
+			opt.MaxRedirects = 0
 		})
 
 		AfterEach(func() {
 			_ = client.ForEachShard(ctx, func(ctx context.Context, client *redis.Client) error {
 				defer GinkgoRecover()
+
+				opt := client.Options()
+				opt.ReadTimeout = time.Second
+				opt.WriteTimeout = time.Second
+
 				Eventually(func() error {
 					return client.Ping(ctx).Err()
 				}, 2*pause).ShouldNot(HaveOccurred())
diff --git a/race_test.go b/race_test.go
index 52181e0..2265366 100644
--- a/race_test.go
+++ b/race_test.go
@@ -2,7 +2,6 @@ package redis_test
 
 import (
 	"bytes"
-	"context"
 	"fmt"
 	"net"
 	"strconv"
@@ -289,26 +288,6 @@ var _ = Describe("races", func() {
 		wg.Wait()
 		Expect(atomic.LoadUint32(&received)).To(Equal(uint32(C * N)))
 	})
-
-	It("should abort on context timeout", func() {
-		opt := redisClusterOptions()
-		client := cluster.newClusterClient(ctx, opt)
-
-		ctx, cancel := context.WithCancel(context.Background())
-
-		wg := performAsync(C, func(_ int) {
-			_, err := client.XRead(ctx, &redis.XReadArgs{
-				Streams: []string{"test", "$"},
-				Block:   1 * time.Second,
-			}).Result()
-			Expect(err).To(HaveOccurred())
-			Expect(err.Error()).To(Or(Equal(context.Canceled.Error()), ContainSubstring("operation was canceled")))
-		})
-
-		time.Sleep(10 * time.Millisecond)
-		cancel()
-		wg.Wait()
-	})
 })
 
 var _ = Describe("cluster races", func() {
diff --git a/redis.go b/redis.go
index 704d9e5..0428149 100644
--- a/redis.go
+++ b/redis.go
@@ -290,27 +290,7 @@ func (c *baseClient) withConn(
 		c.releaseConn(ctx, cn, err)
 	}()
 
-	done := ctx.Done() //nolint:ifshort
-
-	if done == nil {
-		err = fn(ctx, cn)
-		return err
-	}
-
-	errc := make(chan error, 1)
-	go func() { errc <- fn(ctx, cn) }()
-
-	select {
-	case <-done:
-		_ = cn.Close()
-		// Wait for the goroutine to finish and send something.
-		<-errc
-
-		err = ctx.Err()
-		return err
-	case err = <-errc:
-		return err
-	}
+	return fn(ctx, cn)
 }
 
 func (c *baseClient) process(ctx context.Context, cmd Cmder) error {
@@ -416,7 +396,6 @@ func (c *baseClient) generalProcessPipeline(
 ) error {
 	err := c._generalProcessPipeline(ctx, cmds, p)
 	if err != nil {
-		setCmdsErr(cmds, err)
 		return err
 	}
 	return cmdsFirstErr(cmds)
@@ -429,6 +408,7 @@ func (c *baseClient) _generalProcessPipeline(
 	for attempt := 0; attempt <= c.opt.MaxRetries; attempt++ {
 		if attempt > 0 {
 			if err := internal.Sleep(ctx, c.retryBackoff(attempt)); err != nil {
+				setCmdsErr(cmds, err)
 				return err
 			}
 		}
@@ -449,53 +429,61 @@ func (c *baseClient) _generalProcessPipeline(
 func (c *baseClient) pipelineProcessCmds(
 	ctx context.Context, cn *pool.Conn, cmds []Cmder,
 ) (bool, error) {
-	err := cn.WithWriter(ctx, c.opt.WriteTimeout, func(wr *proto.Writer) error {
+	if err := cn.WithWriter(ctx, c.opt.WriteTimeout, func(wr *proto.Writer) error {
 		return writeCmds(wr, cmds)
-	})
-	if err != nil {
+	}); err != nil {
+		setCmdsErr(cmds, err)
 		return true, err
 	}
 
-	err = cn.WithReader(ctx, c.opt.ReadTimeout, func(rd *proto.Reader) error {
+	if err := cn.WithReader(ctx, c.opt.ReadTimeout, func(rd *proto.Reader) error {
 		return pipelineReadCmds(rd, cmds)
-	})
-	return true, err
+	}); err != nil {
+		return true, err
+	}
+
+	return false, nil
 }
 
 func pipelineReadCmds(rd *proto.Reader, cmds []Cmder) error {
-	for _, cmd := range cmds {
+	for i, cmd := range cmds {
 		err := cmd.readReply(rd)
 		cmd.SetErr(err)
 		if err != nil && !isRedisError(err) {
+			setCmdsErr(cmds[i+1:], err)
 			return err
 		}
 	}
-	return nil
+	// Retry errors like "LOADING redis is loading the dataset in memory".
+	return cmds[0].Err()
 }
 
 func (c *baseClient) txPipelineProcessCmds(
 	ctx context.Context, cn *pool.Conn, cmds []Cmder,
 ) (bool, error) {
-	err := cn.WithWriter(ctx, c.opt.WriteTimeout, func(wr *proto.Writer) error {
+	if err := cn.WithWriter(ctx, c.opt.WriteTimeout, func(wr *proto.Writer) error {
 		return writeCmds(wr, cmds)
-	})
-	if err != nil {
+	}); err != nil {
+		setCmdsErr(cmds, err)
 		return true, err
 	}
 
-	err = cn.WithReader(ctx, c.opt.ReadTimeout, func(rd *proto.Reader) error {
+	if err := cn.WithReader(ctx, c.opt.ReadTimeout, func(rd *proto.Reader) error {
 		statusCmd := cmds[0].(*StatusCmd)
 		// Trim multi and exec.
-		cmds = cmds[1 : len(cmds)-1]
+		trimmedCmds := cmds[1 : len(cmds)-1]
 
-		err := txPipelineReadQueued(rd, statusCmd, cmds)
-		if err != nil {
+		if err := txPipelineReadQueued(rd, statusCmd, trimmedCmds); err != nil {
+			setCmdsErr(cmds, err)
 			return err
 		}
 
-		return pipelineReadCmds(rd, cmds)
-	})
-	return false, err
+		return pipelineReadCmds(rd, trimmedCmds)
+	}); err != nil {
+		return false, err
+	}
+
+	return false, nil
 }
 
 func wrapMultiExec(ctx context.Context, cmds []Cmder) []Cmder {