fastdfs集群分片上传问题

环境使用docker部署fastdfs集群,2个tracker，6个storage（分为3组,每组2个）,前后端分离，使用fastdfs-client-java 1.29-SNAPSHOT版本的客户端(自己通过源码编译打成jar包使用)，上传300多M的文件，每次上传10M，第一次调用StorageClient类的upload_appender_file(byte[] file_buff, String file_ext_name,NameValuePair[] meta_list),后面都是调用的StorageClient类的append_file(String group_name, String appender_filename, byte[] file_buff)方法进行续传，每次续传前端会传参已经上传的filesize，后端在续传前会调用get_file_info(String group_name, String remote_filename)方法获取在fastdfs中的filesize，多次续传后get_file_info返回的filesize会越来越比前端传的已上传的filesize小,上传300多M的文件实际上在fastdfs中只有180M(甚至有时候第一个分片上传成功后 get_file_info返回的是null)，程序没有报任何异常，请问是否是storage同步的问题（续传的storage不是上一次上传的storage，当前storage还未从上次上传的storage同步过来全部数据）应如何实现在集群环境下的分片上传？

Start with the reported calls to StorageClient.upload_appender_file, append_file, and get_file_info, then trace how they behave across the Docker-deployed tracker and storage groups. Reproduce the multi-part upload and compare the reported uploaded size with the size returned by FastDFS; done means the cause of the inconsistent size or null result is identified and documented with a reliable cluster upload approach.