pyspark在windows下java.net.SocketException: Connection reset by peer 错误

pyspark在windows加载数据集 训练模型出现 以下错误
java.net.SocketException: Connection reset by peer: socket write error
        at java.net.SocketOutputStream.socketWrite0(Native Method)
        at java.net.SocketOutputStream.socketWrite(SocketOutputStream.java:113)


在stackoverflow.com上找到的解决办法
修改spark/python/pyspark目录下worker.py文件
修改 def process():为以下内容
        def process():
            iterator = deserializer.load_stream(infile)
            serializer.dump_stream(func(split_index, iterator), outfile)
            for obj in iterator:
                pass

然后将pyspark目录打包为zip 替换掉spark/python/lib/目录下对应的pyspark.zip文件

猜你喜欢

转载自strayly.iteye.com/blog/2342892
今日推荐