归档hive历史分区不会减少hdfs存储空间，但是可以有效减轻hadoop namenode的压力，尤其在于小文件比较多的情况下。

$mkdir $HIVE_HOME/auxlib

$ cp /opt/cdh-5.3.6/hadoop-2.5.0/share/hadoop/tools/lib/hadoop-archives-2.5.0-cdh5.3.6.jar /opt/cdh-5.3.6/hive-0.13.1/auxlib/hadoop-archives-2.5.0-cdh5.3.6.jar

hive (chavin)>  set hive.archive.enabled=true;                                    
hive (chavin)> alter table emp archive partition(country=‘china‘,state=‘beijing‘);
intermediate.archived is hdfs://db01:8020/user/hive/warehouse/chavin.db/emp/country=china/state=beijing_INTERMEDIATE_ARCHIVED
intermediate.original is hdfs://db01:8020/user/hive/warehouse/chavin.db/emp/country=china/state=beijing_INTERMEDIATE_ORIGINAL
Creating data.har for hdfs://db01:8020/user/hive/warehouse/chavin.db/emp/country=china/state=beijing
in hdfs://db01:8020/user/hive/warehouse/chavin.db/emp/country=china/state=beijing/.hive-staging_hive_2017-05-24_13-23-51_914_3548751700804069937-1/-ext-10000/partlevel
Please wait... (this may take a while)
Moving hdfs://db01:8020/user/hive/warehouse/chavin.db/emp/country=china/state=beijing/.hive-staging_hive_2017-05-24_13-23-51_914_3548751700804069937-1/-ext-10000/partlevel to hdfs://db01:8020/user/hive/warehouse/chavin.db/emp/country=china/state=beijing_INTERMEDIATE_ARCHIVED
Moving hdfs://db01:8020/user/hive/warehouse/chavin.db/emp/country=china/state=beijing to hdfs://db01:8020/user/hive/warehouse/chavin.db/emp/country=china/state=beijing_INTERMEDIATE_ORIGINAL
Moving hdfs://db01:8020/user/hive/warehouse/chavin.db/emp/country=china/state=beijing_INTERMEDIATE_ARCHIVED to hdfs://db01:8020/user/hive/warehouse/chavin.db/emp/country=china/state=beijing
Moved: ‘hdfs://db01:8020/user/hive/warehouse/chavin.db/emp/country=china/state=beijing_INTERMEDIATE_ORIGINAL‘ to trash at: hdfs://db01:8020/user/hadoop/.Trash/Current
OK
Time taken: 2.17 seconds

可以看到，emp表下分区文件已经被打包成data.har文件了。

hive归档分区