Sqoop

Install Sqoop

Transfer data

su hadoop

• Import data from MySQL into HDFS
  Execute the following command on the Master node of the cluster:

  sqoop import --connect jdbc:mysql://<dburi>/<dbname> --username <username> --password <password> --table <tablename> --check-column <col> --incremental <mode> --last-value <value> --target-dir <hdfs-dir>

  Parameters:

  • dburi: the connection string of a database such as jdbc:mysql://192.168.1.124:3306/. When a connection string includes any parameter, you must enclose the connection string within single quotation marks (') such as jdbc:mysql://192.168.1.124:3306/mydatabase? useUnicode=true
  • dbname: the name of a database such as user.
  • username: the username that is used to log on to a database.
  • password: the password that username with the username.
  • tablename: the name of a MySQL table.
  • col: the name of a column to be queried.
  • mode: specifies how Sqoop determines which rows are new. Valid values: append and lastmodified.
  • value : specifies the maximum value of a column to be checked from the previous import.
  • hdfs-dir: the HDFS directory that you import data into such as/user/hive/result.

  For more information about parameters, see Sqoop import.
• Import data from HDFS into MySQL
  You must create MySQL tables that comply with the data structure of HDFS in advance. Then you can execute the following command on the Master node of a cluster to specify a directory that you import data into.

  sqoop export --connect jdbc:mysql://<dburi>/<dbname> --username <username> --password <password> --table <tablename> --export-dir <hdfs-dir>

  Parameters:

  • dburi: the connection string of a database such as jdbc:mysql://192.168.1.124:3306/. When any parameter is included in a connection string, enclose the connection string within single quotation marks (') such as jdbc:mysql://192.168.1.124:3306/mydatabase? useUnicode=true
  • dbname: the name of a database, such as user.
  • username: the username that is used to log on to a database.
  • password: the password that is associated with the username.
  • tablename: the name of a MySQL table.
  • hdfs-dir: the directory of HDFS from which you import data into MySQL such as /user/hive/result.

  For more information about parameters, see Sqoop export
• Import data from MySQL into Hive
  When you execute the following command on the Master node of a cluster to import data from MySQL, a Hive table will be created as follows.

  sqoop import --connect jdbc:mysql://<dburi>/<dbname> --username <username> --password <password> --table <tablename> --check-column <col> --incremental <mode> --last-value <value> --fields-terminated-by "\t" --lines-terminated-by "\n" --hive-import --target-dir <hdfs-dir> --hive-table <hive-tablename>

  Parameters:

  • dburi: the connection string of a database such as jdbc:mysql://192.168.1.124:3306/. When a connection string includes any parameter, you must enclose the connection string within single quotation marks (') such as jdbc:mysql://192.168.1.124:3306/mydatabase? useUnicode=true
  • dbname: the name of a database such as user.
  • username: the username that is used to log on to a database.
  • password: the password that is associated with the username.
  • tablename: the name of a MySQL table.
  • col: the name of a column to be queried.
  • mode: specifies how Sqoop determine which rows are new. Valid values: append and lastmodified. When you import data into Hive by Sqoop, you cannot use the append mode.
  • value: specifies the maximum value of a column to be queried from the previous import.
  • hdfs-dir: the directory of HDFS from which you import data into MySQL such as /user/hive/result.
  • hive-tablename: the table name of Hive such as xxx.yyy.

  For more information about how to use parameters, see Sqoop import.

• Import data from Hive into MySQL

  You can refer to the previous command that is used to import data from HDFS into MySQL. In addition, you must specify the HDFS directory of Hive tables from which you import data into MySQL.

• Import data from MySQL into OSS
  The process is similar to importing data from MySQL into HDFS, except for the configuration of the target-dir parameter. You can execute the following command on the Master node of a cluster:

  sqoop import --connect jdbc:mysql://<dburi>/<dbname> --username <username> --password <password> --table <tablename> --check-column <col> --incremental <mode> --last-value <value> --target-dir <oss-dir> --temporary-rootdir <oss-tmpdir>

  Notice

  • The endpoint of an OSS host can be: intranet endpoint, Internet endpoint, or VPC endpoint. For a classic network, you must specify the intranet endpoint. For example, the OSS intranet endpoint of the China (Hangzhou) region is oss-cn-hangzhou-internal.aliyuncs.com. For a VPC, you must specify the VPC endpoint. For example, the OSS VPC endpoint of the China (Hangzhou) region is vpc100-oss-cn-hangzhou.aliyuncs.com.
  • Currently, when you import data into OSS, you cannot specify the delete-target-dir parameter. Otherwise, the error message Wrong FS occurs. When you want to overwrite a directory, you can execute the hadoop fs -rm -r osspath command to remove this OSS directory before using Sqoop.

  sqoop import --connect jdbc:mysql://<dburi>/<dbname> --username <username> --password <password> --table <tablename> --check-column <col> --incremental <mode> --last-value <value> --target-dir <oss-dir> --temporary-rootdir <oss-tmpdir>

  Parameters:

  • dburi: the connection string of a database such as jdbc:mysql://192.168.1.124:3306/ . When a connection string includes any parameter, enclose the connection string within single quotation marks (') such as jdbc:mysql://192.168.1.124:3306/mydatabase? useUnicode=true
  • dbname: the name of a database such as user.
  • username: the username that is used to log on to a database.
  • password: the password that is associated with the username.
  • tablename: the name of a MySQL table.
  • col: the name of a column to be queried.
  • mode: used by Sqoop to determine which rows are new rows. Valid values: append and lastmodified.
  • value: specifies the maximum value of a column to be queried from the previous import.
  • oss-dir: the OSS directory that you import data into oss://<accessid>:<accesskey>@<bucketname>.oss-cn-hangzhou-internal.aliyuncs.com/result。
  • oss-tmpdir: the temporary target folder. You must specify this parameter when you specify the append mode. If the destination directory already exists in HDFS, Sqoop will stop to import and overwrite that directory’s contents. If you specify the append mode, Sqoop will import data to a temporary directory and then rename the files to the normal target directory in a manner that does not conflict with the existing filenames in that directory.

  For more information about available parameters, see Sqoop import.

• Import data from OSS into MySQL

  The process is similar to importing data from MySQL to HDFS, except for the configuration of the export-dir parameter. You must create MySQL tables that comply with the data structure of OSS in advance.

  Then you can execute the following command on the Master node of a cluster to specify the directory from which you want to import data.

  sqoop export --connect jdbc:mysql://<dburi>/<dbname> --username <username> --password <password> --table <tablename> --export-dir <oss-dir>

  Parameters:

  • dburi: the connection string of a database such as jdbc:mysql://192.168.1.124:3306/ . When a connection string includes any parameter, you must enclose this connection string within single quotation marks (') such as jdbc:mysql://192.168.1.124:3306/mydatabase? useUnicode=true
  • dbname: the name of a database such as user.
  • username: the username that is used to log on to a database.
  • password: the password that is associated with the username.
  • tablename: the name of a MySQL table.
  • oss-dir: the OSS directory that you import data into such asoss://<accessid>:<accesskey>@<bucketname>.oss-cn-hangzhou-internal.aliyuncs.com/result.
  • oss-tmpdir: the temporary directory that you import data into. You must specify this parameter when you specify the append mode. If the destination directory already exists in HDFS, Sqoop will stop to import and overwrite that directory’s contents. If you specify the append mode, Sqoop will import data to a temporary directory and then rename the files into the normal target directory in a manner that does not conflict with existing filenames in that directory.

  Note The endpoint of an OSS host can be: intranet endpoint, Internet endpoint, or VPC endpoint. For a classic network, you must specify an intranet endpoint. For example, the OSS intranet endpoint of the China (Hangzhou) region is oss-cn-hangzhou-internal.aliyuncs.com. For a VPC, you must specify a VPC endpoint. For example, the OSS VPC endpoint of the China (Hangzhou) region isvpc100-oss-cn-hangzhou.aliyuncs.com.
  For more information about available parameters, see Sqoop export.
• Free-form query imports
  In addition to importing a set of MySQL tables, you can also import the result set of an arbitrary SQL query as follows:

  sqoop import --connect jdbc:mysql://<dburi>/<dbname> --username <username> --password <password> --query <query-sql> --split-by <sp-column> --hive-import --hive-table <hive-tablename> --target-dir <hdfs-dir>

  Parameters:

  • dburi: the connection string of a database, such as jdbc:mysql://192.168.1.124:3306/ . When a connection string includes any parameter, you must enclose this connection string within single quotation marks (') such as jdbc:mysql://192.168.1.124:3306/mydatabase? useUnicode=true
  • dbname: the name of a database such as user.
  • username: the username that is used to log on to a database.
  • password: the password that is associated with the username.
  • query-sql: the query statement such as SELECT * FROM profile WHERE id>1 AND \$CONDITIONS. You must enclose the query statement that ends with AND \$CONDITIONS within double quotation marks (").
  • sp-column: specifies the name of a column to be split. In general, the value of this parameter is the primary key of the MySQL table.
  • hdfs-dir: the directory of HDFS from which you import data into MySQL such as /user/hive/result.
  • hive-tablename: the name of a table that is used to import data to Hive such as xxx.yyy.

  For more information about available parameters, see Sqoop query import.