Hive - 外部(动态)分区表

Question

我在MySQL中有一个表.nas_comps.

select comp_code, count(leg_id) from nas_comps_01012011_31012011 n group by comp_code;
comp_code     count(leg_id)
'J'           20640
'Y'           39680

首先,我使用Sqoop将数据导入HDFSHadoop版本1.0.2):

sqoop import --connect jdbc:mysql://172.25.37.135/pros_olap2 \
--username hadoopranch \
--password hadoopranch \
--query "select * from nas_comps where dep_date between '2011-01-01' and '2011-01-10' AND \$CONDITIONS" \
-m 1 \
--target-dir /pros/olap2/dataimports/nas_comps

然后,我创建了一个外部的分区Hive表:

/*shows the partitions on 'describe' but not 'show partitions'*/
create external table  nas_comps(DS_NAME string,DEP_DATE string,
                                 CRR_CODE string,FLIGHT_NO string,ORGN string,
                                 DSTN string,PHYSICAL_CAP int,ADJUSTED_CAP int,
                                 CLOSED_CAP int)
PARTITIONED BY (LEG_ID int, month INT, COMP_CODE string)
location '/pros/olap2/dataimports/nas_comps'

分区列在描述时显示:

hive> describe extended nas_comps;
OK
ds_name string
dep_date        string
crr_code        string
flight_no       string
orgn    string
dstn    string
physical_cap    int
adjusted_cap    int
closed_cap      int
leg_id  int
month   int
comp_code       string

Detailed Table Information      Table(tableName:nas_comps, dbName:pros_olap2_optim, 
owner:hadoopranch, createTime:1374849456, lastAccessTime:0, retention:0, 
sd:StorageDescriptor(cols:[FieldSchema(name:ds_name, type:string, comment:null), 
FieldSchema(name:dep_date, type:string, comment:null), FieldSchema(name:crr_code, 
type:string, comment:null), FieldSchema(name:flight_no, type:string, comment:null), 
FieldSchema(name:orgn, type:string, comment:null), FieldSchema(name:dstn, type:string, 
comment:null), FieldSchema(name:physical_cap, type:int, comment:null), 
FieldSchema(name:adjusted_cap, type:int, comment:null), FieldSchema(name:closed_cap, 
type:int, comment:null), FieldSchema(name:leg_id, type:int, comment:null), 
FieldSchema(name:month, type:int, comment:null), FieldSchema(name:comp_code, type:string, 
comment:null)], location:hdfs://172.25.37.21:54300/pros/olap2/dataimports/nas_comps, 
inputFormat:org.apache.hadoop.mapred.TextInputFormat, 
outputFormat:org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat, compressed:false, 
numBuckets:-1, serdeInfo:SerDeInfo(name:null, 
serializationLib:org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe, parameters:
{serialization.format=1}), bucketCols:[], sortCols:[], parameters:{}), partitionKeys:
[FieldSchema(name:leg_id, type:int, comment:null), FieldSchema(name:month, type:int,
comment:null), FieldSchema(name:comp_code, type:string, comment:null)], 
parameters:{EXTERNAL=TRUE, transient_lastDdlTime=1374849456}, viewOriginalText:null, 
viewExpandedText:null, tableType:EXTERNAL_TABLE)

但我不确定是否创建了分区,因为:

hive> show partitions nas_comps;
OK
Time taken: 0.599 seconds


select count(1) from nas_comps;

返回0条记录

如何使用动态分区创建外部Hive表？

Answer 1

Hive不会以这种方式为您创建分区.
只需创建一个由所需分区键分区的表,然后insert overwrite table从外部表执行到新的分区表(设置hive.exec.dynamic.partition=true和hive.exec.dynamic.partition.mode=nonstrict).

如果必须保持外部表分区,则必须手动创建目录(每个分区名称应为1个目录PARTION_KEY=VALUE)然后使用MSCK REPAIR TABLE table_name; 命令

Answer 2

动态分区

在将记录插入hive表期间动态添加分区.

1. 仅支持insert语句.
2. load data声明不支持.
3. 在将数据插入配置单元表之前需要启用动态分区设置. hive.exec.dynamic.partition.mode=nonstrict默认值是strict  hive.exec.dynamic.partition=true默认值是false.

动态分区查询

SET hive.exec.dynamic.partition.mode=nonstrict;
SET hive.exec.dynamic.partition=true;
INSERT INTO table_name PARTITION (loaded_date)
select * from table_name1 where loaded_date = 20151217

这loaded_date = 20151217是分区及其值.

限制:

1. 动态分区仅适用于上述语句.
2. 它将动态创建分区按照它从选择数据loaded_date从列table_name1;

如果您的情况与上述标准不符,则:

首先创建一个分区表,然后执行以下操作:

ALTER TABLE table_name ADD PARTITION (DS_NAME='partname1',DATE='partname2'); 

或者请使用此 链接创建动态分区.