br: configure the httpclient for external storage (#46040) (#46140)

close #46011

br/pkg/storage/BUILD.bazel

@@ -67,11 +67,12 @@ go_test(
         "memstore_test.go",
         "parse_test.go",
         "s3_test.go",
+        "storage_test.go",
         "writer_test.go",
     ],
     embed = [":storage"],
     flaky = True,
-    shard_count = 43,
+    shard_count = 47,
     deps = [
         "//br/pkg/mock",
         "@com_github_aws_aws_sdk_go//aws",

br/pkg/storage/azblob.go

@@ -107,10 +107,12 @@ type sharedKeyClientBuilder struct {
 	cred        *azblob.SharedKeyCredential
 	accountName string
 	serviceURL  string
+
+	clientOptions *azblob.ClientOptions
 }
 
 func (b *sharedKeyClientBuilder) GetServiceClient() (azblob.ServiceClient, error) {
-	return azblob.NewServiceClientWithSharedKey(b.serviceURL, b.cred, getDefaultClientOptions())
+	return azblob.NewServiceClientWithSharedKey(b.serviceURL, b.cred, b.clientOptions)
 }
 
 func (b *sharedKeyClientBuilder) GetAccountName() string {
@@ -122,10 +124,12 @@ type tokenClientBuilder struct {
 	cred        *azidentity.ClientSecretCredential
 	accountName string
 	serviceURL  string
+
+	clientOptions *azblob.ClientOptions
 }
 
 func (b *tokenClientBuilder) GetServiceClient() (azblob.ServiceClient, error) {
-	return azblob.NewServiceClient(b.serviceURL, b.cred, getDefaultClientOptions())
+	return azblob.NewServiceClient(b.serviceURL, b.cred, b.clientOptions)
 }
 
 func (b *tokenClientBuilder) GetAccountName() string {
@@ -144,6 +148,11 @@ func getAzureServiceClientBuilder(options *backuppb.AzureBlobStorage, opts *Exte
 		return nil, errors.New("bucket(container) cannot be empty to access azure blob storage")
 	}
 
+	clientOptions := getDefaultClientOptions()
+	if opts != nil && opts.HTTPClient != nil {
+		clientOptions.Transporter = opts.HTTPClient
+	}
+
 	if len(options.AccountName) > 0 && len(options.SharedKey) > 0 {
 		serviceURL := options.Endpoint
 		if len(serviceURL) == 0 {
@@ -157,6 +166,8 @@ func getAzureServiceClientBuilder(options *backuppb.AzureBlobStorage, opts *Exte
 			cred,
 			options.AccountName,
 			serviceURL,
+
+			clientOptions,
 		}, nil
 	}
 
@@ -185,6 +196,8 @@ func getAzureServiceClientBuilder(options *backuppb.AzureBlobStorage, opts *Exte
 				cred,
 				accountName,
 				serviceURL,
+
+				clientOptions,
 			}, nil
 		}
 		log.Warn("Failed to get azure token credential but environment variables exist, try to use shared key.", zap.String("tenantId", tenantID), zap.String("clientId", clientID), zap.String("clientSecret", "?"))
@@ -212,6 +225,8 @@ func getAzureServiceClientBuilder(options *backuppb.AzureBlobStorage, opts *Exte
 		cred,
 		accountName,
 		serviceURL,
+
+		clientOptions,
 	}, nil
 }
 

br/pkg/storage/storage.go

@@ -27,6 +27,8 @@ const (
 	GetObject Permission = "GetObject"
 	// PutObject represents PutObject permission
 	PutObject Permission = "PutObject"
+
+	DefaultRequestConcurrency uint = 128
 )
 
 // WalkOption is the option of storage.WalkDir.
@@ -187,3 +189,19 @@ func New(ctx context.Context, backend *backuppb.StorageBackend, opts *ExternalSt
 		return nil, errors.Annotatef(berrors.ErrStorageInvalidConfig, "storage %T is not supported yet", backend)
 	}
 }
+
+// Different from `http.DefaultTransport`, set the `MaxIdleConns` and `MaxIdleConnsPerHost`
+// to the actual request concurrency to reuse tcp connection as much as possible.
+func GetDefaultHttpClient(concurrency uint) *http.Client {
+	transport, _ := CloneDefaultHttpTransport()
+	transport.MaxIdleConns = int(concurrency)
+	transport.MaxIdleConnsPerHost = int(concurrency)
+	return &http.Client{
+		Transport: transport,
+	}
+}
+
+func CloneDefaultHttpTransport() (*http.Transport, bool) {
+	transport, ok := http.DefaultTransport.(*http.Transport)
+	return transport.Clone(), ok
+}

br/pkg/storage/storage_test.go

@@ -0,0 +1,26 @@
+// Copyright 2022 PingCAP, Inc. Licensed under Apache-2.0.
+
+package storage_test
+
+import (
+	"net/http"
+	"testing"
+
+	"github.com/pingcap/tidb/br/pkg/storage"
+	"github.com/stretchr/testify/require"
+)
+
+func TestDefaultHttpTransport(t *testing.T) {
+	transport, ok := storage.CloneDefaultHttpTransport()
+	require.True(t, ok)
+	require.True(t, transport.MaxConnsPerHost == 0)
+	require.True(t, transport.MaxIdleConns > 0)
+}
+
+func TestDefaultHttpClient(t *testing.T) {
+	var concurrency uint = 128
+	transport, ok := storage.GetDefaultHttpClient(concurrency).Transport.(*http.Transport)
+	require.True(t, ok)
+	require.Equal(t, int(concurrency), transport.MaxIdleConnsPerHost)
+	require.Equal(t, int(concurrency), transport.MaxIdleConns)
+}

br/pkg/task/stream.go

@@ -132,6 +132,7 @@ func (cfg *StreamConfig) makeStorage(ctx context.Context) (storage.ExternalStora
 	opts := storage.ExternalStorageOptions{
 		NoCredentials:   cfg.NoCreds,
 		SendCredentials: cfg.SendCreds,
+		HTTPClient:      storage.GetDefaultHttpClient(cfg.MetadataDownloadBatchSize),
 	}
 	storage, err := storage.New(ctx, u, &opts)
 	if err != nil {
@@ -1333,6 +1334,7 @@ func createRestoreClient(ctx context.Context, g glue.Glue, cfg *RestoreConfig, m
 	opts := storage.ExternalStorageOptions{
 		NoCredentials:   cfg.NoCreds,
 		SendCredentials: cfg.SendCreds,
+		HTTPClient:      storage.GetDefaultHttpClient(cfg.MetadataDownloadBatchSize),
 	}
 	if err = client.SetStorage(ctx, u, &opts); err != nil {
 		return nil, errors.Trace(err)