@Fan WeiWen Could you paste the complete DDL of the table which you are querying? ... View more

I have solved this problem ... View more

Who can give me advice ? now i tcpkill port ... View more

@Fan WeiWen How are the new file size (with simple insert with no aggregation) comparable to the files under source table? and what are the block size you have defined on HDFS and ORC levels? Anyway, you should be able to control file size produced by mappers, when you run without reduce functions. try (assuming tez engine): set hive.hadoop.supports.splittable.combineinputformat=true; set hive.merge.tezfiles=true; -- this is to merge files as part of the job set tez.grouping.max-size=268435456; set tez.grouping.min-size=134217728; -- this is to minimize number of mappers created for a job ... View more

@Fan WeiWen Usually ORC output format is being used along for Hive tables ORC formatted. But if you really want to use ORC output format from pure M/R job, you need to provide a schema. ORC is columnar storage format. In order to build those columns, output formatter needs to know how it should organize the data. First thing I noted from provided command line is missing attribute "orc.mapred.output.schema". For more information please refer to https://orc.apache.org/docs/mapreduce.html ... View more