diff --git a/api/v1/api_uploads.go b/api/v1/api_uploads.go
index 6f042ef78061ad13cf9fd993ec631c3c13ce0179..b04d1c1405c52a9fddec2b3197dd25fa0c3a2094 100644
--- a/api/v1/api_uploads.go
+++ b/api/v1/api_uploads.go
@@ -120,30 +120,42 @@ func newFlowJSFile(size uint64, numChunks uint, job *importer.Job) (*FlowJSFile,
 }
 
 func (f *FlowJSFile) Read(p []byte) (n int, err error) {
-	if f.readOffset >= len(f.chunks) {
-		// we already read all the chunks
-		return 0, io.EOF
-	}
+	for {
+		if f.readOffset >= len(f.chunks) {
+			// we already read all the chunks
+			return 0, io.EOF
+		}
 
-	// also wait for jobs context !!!?
-	select {
-	case <-f.job.Ctx.Done():
-		return 0, f.job.Ctx.Err()
-	case <-f.chunks[f.readOffset].completed:
-	}
-	if n, err = f.chunks[f.readOffset].Read(p); err == nil || err == io.EOF {
-		return
-	}
+		// wait for chunk to be completed
+		select {
+		case <-f.job.Ctx.Done():
+			return 0, f.job.Ctx.Err()
+		case <-f.chunks[f.readOffset].completed:
+		}
 
-	// we are done with this chunk, clean it up...
-	if err = f.chunks[f.readOffset].Cleanup(); err != nil {
-		return
+		if n, err = f.chunks[f.readOffset].Read(p); err == nil || err == io.EOF {
+			if n > 0 {
+				err = nil
+				return
+			}
+			// we are at the end of the chunk but haven't read anything yet...
+			// .. move on to the next chunk
+			f.readOffset++
+			continue
+		}
+
+		// we are done with this chunk, clean it up...
+		if err = f.chunks[f.readOffset].Cleanup(); err != nil {
+			return
+		}
+		// ...and proceed to the next
+		f.readOffset++
+		if n > 0 {
+			// if this was the last chunk the next call to Read() will return (0, io.EOF)
+			// for now we will return (n, nil)
+			return
+		}
 	}
-	// ...and proceed to the next
-	f.readOffset++
-	// if this was the last chunk the next call to Read() will return (0, io.EOF)
-	// for now we will return (n, nil)
-	return
 }
 
 func (api *API) UploadFileFlowJS() http.Handler {