Merged

.github/workflows/tools.yml

@@ -46,12 +46,18 @@ jobs:
         make tool-bootstrap tool-transit
         mkdir boot-tools
         mv bootstrap transit boot-tools/
+        sha256sum boot-tools/bootstrap > boot-tools/bootstrap.sha256sum
+        cat boot-tools/bootstrap.sha256sum
+        sha256sum boot-tools/transit > boot-tools/transit.sha256sum
+        cat boot-tools/transit.sha256sum
         tar -czf boot-tools.tar ./boot-tools/
         gsutil cp boot-tools.tar gs://flow-genesis-bootstrap/tools/${{ inputs.tag }}/boot-tools.tar
     - name: Build and upload util
       run: |
         make tool-util
-        tar -czf util.tar util
+        sha256sum util > util.sha256sum
+        cat util.sha256sum
+        tar -czf util.tar util util.sha256sum
         gsutil cp util.tar gs://flow-genesis-bootstrap/tools/${{ inputs.tag }}/util.tar
     - name: Promote boot-tools
       run: |

cmd/bootstrap/cmd/pull.go

@@ -1,21 +1,26 @@
 package cmd
 
 import (
+	"bytes"
 	"context"
 	"fmt"
 	"path/filepath"
 	"strings"
+	"sync"
 	"time"
 
 	"github.com/spf13/cobra"
+	"golang.org/x/sync/semaphore"
 
 	"github.com/onflow/flow-go/cmd"
 	"github.com/onflow/flow-go/cmd/bootstrap/gcs"
+	"github.com/onflow/flow-go/cmd/bootstrap/utils"
 )
 
 var (
-	flagNetwork    string
-	flagBucketName string
+	flagNetwork     string
+	flagBucketName  string
+	flagConcurrency int64
 )
 
 // pullCmd represents a command to pull parnter node details from the google
@@ -37,6 +42,7 @@ func addPullCmdFlags() {
 	cmd.MarkFlagRequired(pullCmd, "network")
 
 	pullCmd.Flags().StringVar(&flagBucketName, "bucket", "flow-genesis-bootstrap", "google bucket name")
+	pullCmd.Flags().Int64Var(&flagConcurrency, "concurrency", 2, "concurrency limit")
 }
 
 // pull partner node info from google bucket
@@ -62,15 +68,35 @@ func pull(cmd *cobra.Command, args []string) {
 	}
 	log.Info().Msgf("found %d files in google bucket", len(files))
 
+	sem := semaphore.NewWeighted(flagConcurrency)
+	wait := sync.WaitGroup{}
 	for _, file := range files {
-		if strings.Contains(file, "node-info.pub") {
-			fullOutpath := filepath.Join(flagOutdir, file)
-			log.Printf("downloading %s", file)
-
-			err = bucket.DownloadFile(ctx, client, fullOutpath, file)
-			if err != nil {
-				log.Error().Msgf("error trying download google bucket file: %v", err)
+		wait.Add(1)
+		go func(file gcs.GCSFile) {
+			_ = sem.Acquire(ctx, 1)
+			defer func() {
+				sem.Release(1)
+				wait.Done()
+			}()
+
+			if strings.Contains(file.Name, "node-info.pub") {
+				fullOutpath := filepath.Join(flagOutdir, file.Name)
+
+				fmd5 := utils.CalcMd5(fullOutpath)
+				// only skip files that have an MD5 hash
+				if file.MD5 != nil && bytes.Equal(fmd5, file.MD5) {
+					log.Printf("skipping %s", file)
+					return
+				}
+
+				log.Printf("downloading %s", file)
+				err = bucket.DownloadFile(ctx, client, fullOutpath, file.Name)
+				if err != nil {
+					log.Error().Msgf("error trying download google bucket file: %v", err)
+				}
 			}
-		}
+		}(file)
 	}
+
+	wait.Wait()
 }

cmd/bootstrap/gcs/gcs.go

@@ -33,14 +33,19 @@ func (g *googleBucket) NewClient(ctx context.Context) (*storage.Client, error) {
 	return client, nil
 }
 
+type GCSFile struct {
+	Name string
+	MD5  []byte
+}
+
 // GetFiles returns a list of file names within the Google bucket
-func (g *googleBucket) GetFiles(ctx context.Context, client *storage.Client, prefix, delimiter string) ([]string, error) {
+func (g *googleBucket) GetFiles(ctx context.Context, client *storage.Client, prefix, delimiter string) ([]GCSFile, error) {
 	it := client.Bucket(g.Name).Objects(ctx, &storage.Query{
 		Prefix:    prefix,
 		Delimiter: delimiter,
 	})
 
-	var files []string
+	var files []GCSFile
 	for {
 		attrs, err := it.Next()
 		if err == iterator.Done {
@@ -50,7 +55,10 @@ func (g *googleBucket) GetFiles(ctx context.Context, client *storage.Client, pre
 			return nil, err
 		}
 
-		files = append(files, attrs.Name)
+		files = append(files, GCSFile{
+			Name: attrs.Name,
+			MD5:  attrs.MD5,
+		})
 	}
 
 	return files, nil

cmd/bootstrap/transit/cmd/flags.go

@@ -9,6 +9,7 @@ var (
 	flagAccessAddress string
 	flagNodeRole      string
 	flagTimeout       time.Duration
+	flagConcurrency   int64
 
 	flagWrapID   string // wrap ID
 	flagVoteFile string

cmd/bootstrap/transit/cmd/pull.go

@@ -1,16 +1,20 @@
 package cmd
 
 import (
+	"bytes"
 	"context"
 	"fmt"
 	"io/fs"
 	"path/filepath"
 	"strings"
+	"sync"
 	"time"
 
 	"github.com/spf13/cobra"
+	"golang.org/x/sync/semaphore"
 
 	"github.com/onflow/flow-go/cmd/bootstrap/gcs"
+	"github.com/onflow/flow-go/cmd/bootstrap/utils"
 	model "github.com/onflow/flow-go/model/bootstrap"
 	"github.com/onflow/flow-go/model/flow"
 )
@@ -32,6 +36,7 @@ func addPullCmdFlags() {
 	pullCmd.Flags().StringVarP(&flagToken, "token", "t", "", "token provided by the Flow team to access the Transit server")
 	pullCmd.Flags().StringVarP(&flagNodeRole, "role", "r", "", `node role (can be "collection", "consensus", "execution", "verification" or "access")`)
 	pullCmd.Flags().DurationVar(&flagTimeout, "timeout", time.Second*300, `timeout for pull`)
+	pullCmd.Flags().Int64Var(&flagConcurrency, "concurrency", 2, `concurrency limit for pull`)
 
 	_ = pullCmd.MarkFlagRequired("token")
 	_ = pullCmd.MarkFlagRequired("role")
@@ -78,17 +83,34 @@ func pull(cmd *cobra.Command, args []string) {
 	}
 	log.Info().Msgf("found %d files in Google Bucket", len(files))
 
-	// download found files
+	sem := semaphore.NewWeighted(flagConcurrency)
+	wait := sync.WaitGroup{}
 	for _, file := range files {
-		fullOutpath := filepath.Join(flagBootDir, "public-root-information", filepath.Base(file))
-
-		log.Info().Str("source", file).Str("dest", fullOutpath).Msgf("downloading file from transit servers")
-		err = bucket.DownloadFile(ctx, client, fullOutpath, file)
-		if err != nil {
-			log.Fatal().Err(err).Msgf("could not download google bucket file")
-		}
+		wait.Add(1)
+		go func(file gcs.GCSFile) {
+			_ = sem.Acquire(ctx, 1)
+			defer func() {
+				sem.Release(1)
+				wait.Done()
+			}()
+
+			fullOutpath := filepath.Join(flagBootDir, "public-root-information", filepath.Base(file.Name))
+			fmd5 := utils.CalcMd5(fullOutpath)
+			// only skip files that have an MD5 hash
+			if file.MD5 != nil && bytes.Equal(fmd5, file.MD5) {
+				log.Info().Str("source", file.Name).Str("dest", fullOutpath).Msgf("skipping existing file from transit servers")
+				return
+			}
+			log.Info().Str("source", file.Name).Str("dest", fullOutpath).Msgf("downloading file from transit servers")
+			err = bucket.DownloadFile(ctx, client, fullOutpath, file.Name)
+			if err != nil {
+				log.Fatal().Err(err).Msgf("could not download google bucket file")
+			}
+		}(file)
 	}
 
+	wait.Wait()
+
 	// download any extra files specific to node role
 	extraFiles := getAdditionalFilesToDownload(role, nodeID)
 	for _, file := range extraFiles {

cmd/bootstrap/utils/md5.go

@@ -0,0 +1,25 @@
+package utils
+
+// The google storage API only provides md5 and crc32 hence overriding the linter flag for md5
+// #nosec
+import (
+	"crypto/md5"
+	"io"
+	"os"
+)
+
+func CalcMd5(outpath string) []byte {
+	f, err := os.Open(outpath)
+	if err != nil {
+		return nil
+	}
+	defer f.Close()
+
+	// #nosec
+	h := md5.New()
+	if _, err := io.Copy(h, f); err != nil {
+		return nil
+	}
+
+	return h.Sum(nil)
+}

cmd/util/cmd/check-storage/cmd.go

@@ -11,7 +11,6 @@ import (
 	"github.com/onflow/cadence/runtime"
 	"github.com/onflow/cadence/runtime/common"
 
-	"github.com/onflow/flow-go/cmd/util/ledger/migrations"
 	"github.com/onflow/flow-go/cmd/util/ledger/reporters"
 	"github.com/onflow/flow-go/cmd/util/ledger/util"
 	"github.com/onflow/flow-go/cmd/util/ledger/util/registers"
@@ -282,6 +281,9 @@ func checkStorageHealth(
 
 		err = registersByAccount.ForEachAccount(
 			func(accountRegisters *registers.AccountRegisters) error {
+				if slices.Contains(acctsToSkip, accountRegisters.Owner()) {
+					return nil
+				}
 				jobs <- job{accountRegisters: accountRegisters}
 				return nil
 			})
@@ -323,7 +325,7 @@ func checkAccountStorageHealth(accountRegisters *registers.AccountRegisters, nWo
 	ledger := &registers.ReadOnlyLedger{Registers: accountRegisters}
 	storage := runtime.NewStorage(ledger, nil)
 
-	err = util.CheckStorageHealth(address, storage, accountRegisters, migrations.AllStorageMapDomains, nWorkers)
+	err = util.CheckStorageHealth(address, storage, accountRegisters, util.StorageMapDomains, nWorkers)
 	if err != nil {
 		issues = append(
 			issues,