Hive数据迁移4（内外部表）

版权声明：本文为博主原创文章，未经博主允许不得转载。https://www.jianshu.com/p/34cc4e2cf181

#date:2019-05-13

#author:Wang Kuan

dbname=$1

#categorize the table type

cat /home/tablename | while read table;

do

  hive -e "use ${dbname};desc $table;" 2>&1 |grep 'Table not found'

  if [ $> -ne 0 ]; then

    hive -e "desc formatted $table;" 2>&1 |grep 'EXTERNAL_TABLE'

    if [ $> -ne 0 ]; then

        echo $table " is internal table"

        echo $table >> internal_table.txt

    else

        echo $table " is external table"

        echo $table >> external_table.txt

    fi

  else

    echo $table " is not exist"

    echo $table >> table_not_exits.txt

  fi

done

#repair the external table

cat /home/external_table.txt | while read externaltable;

do

    hive -e "msck repair table $externaltable";

    echo $externaltable " have been repaired!"

done

#repair the table_not_exist

cat /home/table_not_exist.txt | while read table_not_exist;

do

  beeline -u jdbc:hive2://sourceip:10010 -e "use ${dbname};show tables;"|grep -i $table_not_exist

  if [ $? -ne 0 ];then

    echo $table_not_exist " is not exist in source cluster"

  else

    beeline -u jdbc:hive2://ip:10010 --silent=true --outputformat=csv2 -e "show create table $table_not_exist;">> TableDDL.txt

  fi

done

#repair the internal table

start_day="20190303"

end_day="20190509"

cat /home/internal_table.txt | while read internaltable

do

  echo $internaltable "begin to add partition"

  while [[ $start_date < $end_date ]]

  do

    hive -e "use ${dbname};alter table $internaltable add partition(dt=$start_day)";

    start_date=`date -d "+1 day $start_date" +%Y%m%d`

  done

done

对于内部表来说，drop表HDFS上数据也会删除，而外部表不会。

因此如果新建内部表，需要先将其在HDFS上对应的目录临时移动到别的位置，否则创建表时会将原始目录文件清空。待创建好内部表之后，再移回HDFS文件，添加分区即可。