package weed_server import ( "fmt" "os" "strings" "time" "github.com/seaweedfs/seaweedfs/weed/operation" "google.golang.org/grpc" "github.com/seaweedfs/seaweedfs/weed/pb" "github.com/seaweedfs/seaweedfs/weed/security" "github.com/seaweedfs/seaweedfs/weed/storage/backend" "github.com/seaweedfs/seaweedfs/weed/storage/erasure_coding" "golang.org/x/net/context" "github.com/seaweedfs/seaweedfs/weed/glog" "github.com/seaweedfs/seaweedfs/weed/pb/master_pb" "github.com/seaweedfs/seaweedfs/weed/util" ) func (vs *VolumeServer) GetMaster(ctx context.Context) pb.ServerAddress { return vs.currentMaster } func (vs *VolumeServer) checkWithMaster() (err error) { for { for _, master := range vs.SeedMasterNodes { err = operation.WithMasterServerClient(false, master, vs.grpcDialOption, func(masterClient master_pb.SeaweedClient) error { resp, err := masterClient.GetMasterConfiguration(context.Background(), &master_pb.GetMasterConfigurationRequest{}) if err != nil { return fmt.Errorf("get master %s configuration: %v", master, err) } vs.metricsAddress, vs.metricsIntervalSec = resp.MetricsAddress, int(resp.MetricsIntervalSeconds) backend.LoadFromPbStorageBackends(resp.StorageBackends) return nil }) if err == nil { return } else { glog.V(0).Infof("checkWithMaster %s: %v", master, err) } } time.Sleep(1790 * time.Millisecond) } } func (vs *VolumeServer) heartbeat() { glog.V(0).Infof("Volume server start with seed master nodes: %v", vs.SeedMasterNodes) vs.store.SetDataCenter(vs.dataCenter) vs.store.SetRack(vs.rack) grpcDialOption := security.LoadClientTLS(util.GetViper(), "grpc.volume") var err error var newLeader pb.ServerAddress duplicateRetryCount := 0 for vs.isHeartbeating { for _, master := range vs.SeedMasterNodes { if newLeader != "" { // the new leader may actually is the same master // need to wait a bit before adding itself time.Sleep(3 * time.Second) master = newLeader } vs.store.MasterAddress = master newLeader, err = vs.doHeartbeatWithRetry(master, grpcDialOption, time.Duration(vs.pulseSeconds)*time.Second, duplicateRetryCount) if err != nil { glog.V(0).Infof("heartbeat to %s error: %v", master, err) // Check if this is a duplicate UUID retry error if strings.Contains(err.Error(), "duplicate UUIDs detected, retrying connection") { duplicateRetryCount++ retryDelay := time.Duration(1<<(duplicateRetryCount-1)) * 2 * time.Second // exponential backoff: 2s, 4s, 8s glog.V(0).Infof("Waiting %v before retrying due to duplicate UUID detection...", retryDelay) time.Sleep(retryDelay) } else { // Regular error, reset duplicate retry count duplicateRetryCount = 0 time.Sleep(time.Duration(vs.pulseSeconds) * time.Second) } newLeader = "" vs.store.MasterAddress = "" } else { // Successful connection, reset retry count duplicateRetryCount = 0 } if !vs.isHeartbeating { break } } } } func (vs *VolumeServer) StopHeartbeat() (isAlreadyStopping bool) { if !vs.isHeartbeating { return true } vs.isHeartbeating = false close(vs.stopChan) return false } func (vs *VolumeServer) doHeartbeat(masterAddress pb.ServerAddress, grpcDialOption grpc.DialOption, sleepInterval time.Duration) (newLeader pb.ServerAddress, err error) { return vs.doHeartbeatWithRetry(masterAddress, grpcDialOption, sleepInterval, 0) } func (vs *VolumeServer) doHeartbeatWithRetry(masterAddress pb.ServerAddress, grpcDialOption grpc.DialOption, sleepInterval time.Duration, duplicateRetryCount int) (newLeader pb.ServerAddress, err error) { ctx, cancel := context.WithCancel(context.Background()) defer cancel() grpcConnection, err := pb.GrpcDial(ctx, masterAddress.ToGrpcAddress(), false, grpcDialOption) if err != nil { return "", fmt.Errorf("fail to dial %s : %v", masterAddress, err) } defer grpcConnection.Close() client := master_pb.NewSeaweedClient(grpcConnection) stream, err := client.SendHeartbeat(ctx) if err != nil { glog.V(0).Infof("SendHeartbeat to %s: %v", masterAddress, err) return "", err } glog.V(0).Infof("Heartbeat to: %v", masterAddress) vs.currentMaster = masterAddress doneChan := make(chan error, 1) go func() { for { in, err := stream.Recv() if err != nil { doneChan <- err return } if len(in.DuplicatedUuids) > 0 { var duplicateDir []string for _, loc := range vs.store.Locations { for _, uuid := range in.DuplicatedUuids { if uuid == loc.DirectoryUuid { duplicateDir = append(duplicateDir, loc.Directory) } } } // Implement retry logic for potential race conditions const maxRetries = 3 if duplicateRetryCount < maxRetries { retryDelay := time.Duration(1<