hive又遇到了问题:
原来的mysql中sql语句为
SELECT DATEBUF, CITYID, CHANNELID FROM STAT_CVCHNEW WHERE DATEBUF>=20110517 AND DATEBUF<=20110519 GROUP BY DATEBUF, CITYID;
得到的结果应该是是按DATEBUF和CITYID分组,然后每组对应CHANNELID的第一个值
但在hive中报错
FAILED: SemanticException [Error 10025]: Line 1:20 Expression not in GROUP BY key 'uc'
然后google到http://*.com/questions/5746687/hive-expression-not-in-group-by-key中这样解决
用hive的udaf函数:collect_set()
改为
SELECT DATEBUF, CITYID, collect_set(CHANNELID) FROM PWD_STAT_CVCHNEW WHERE DATEBUF=20130810 and cityid<3 GROUP BY DATEBUF, CITYID;返回的是每一组datebuf和cityid相应的channelid的所有的一个集合
改为
SELECT DATEBUF, CITYID, collect_set(CHANNELID)[0] FROM PWD_STAT_CVCHNEW WHERE DATEBUF=20130810 and cityid<3 GROUP BY DATEBUF, CITYID;ok
但返回的结果channelid是以hdfs中存储的顺序的第一个,
其实这个在一般情况下好像用不到