set datanucleus.cache.collections=false; set hive.stats.autogather=true; set hive.merge.mapfiles=false; set hive.merge.mapredfiles=false; set hive.map.aggr=true; -- SORT_QUERY_RESULTS create table tmptable(key string, value string); EXPLAIN INSERT OVERWRITE TABLE tmptable SELECT unionsrc.key, unionsrc.value FROM (SELECT 'tst1' AS key, cast(count(1) AS string) AS value FROM src s1 UNION ALL SELECT s2.key AS key, s2.value AS value FROM src1 s2) unionsrc; INSERT OVERWRITE TABLE tmptable SELECT unionsrc.key, unionsrc.value FROM (SELECT 'tst1' AS key, cast(count(1) AS string) AS value FROM src s1 UNION ALL SELECT s2.key AS key, s2.value AS value FROM src1 s2) unionsrc; SELECT * FROM tmptable x SORT BY x.key, x.value; DESCRIBE FORMATTED tmptable; -- Load a file into a existing table -- Some stats (numFiles, totalSize) should be updated correctly -- Some other stats (numRows, rawDataSize) should be cleared load data local inpath '../../data/files/srcbucket20.txt' INTO TABLE tmptable; DESCRIBE FORMATTED tmptable;