Command-line options

OBLOADER uses command-line options to specify the information required for import. For more information about options and examples, see Options and Examples.

File matching rules

When you import data, OBLOADER searches for files that match the specified command-line parameters. The default rule for OBLOADER to search for data files to import is ${table_name}.${seq}.${file_type}, where ${seq} is optional. For example, if you specify --csv --table 'table1', OBLOADER will search for files that match the rule table1.${seq}.csv recursively.

You can also use wildcards to specify the files to be imported. For example, if you specify --csv --table '*', OBLOADER will search for all CSV files and split the filenames according to the ${table_name}.${seq}.${file_type} pattern to extract the table names, thereby enabling multi-table import.

If the filenames of the files to be imported are unrelated to the table names, you can use the --file-regular-expression option to specify the matching rule. For example, if the filenames are part-00123-bd461e82-c911-48ce-bfa0-ec1****.c000.snappy.orc, you can use the --orc --table 'table2' --file-regular-expression '.*.orc' command to match all files that end with '.orc' and import them to the specified table named table2.

OBLOADER V4.3.2 and later versions support importing multiple tables in a database by specifying matching rules. You can use the group names (including schema and table) to specify the database name and table names. If only the schema group name is specified, the --table option must specify a single table or a database with only one table, otherwise OBLOADER will report an error and exit. The following rules are supported.

• ${schema}.${table}.csv: ^(?<schema>[^.]+)(\.)(?<table>[^.]+)(\.)csv$

• ${table}.csv: ^(?<table>[^.]+)(\.)csv$

• ${schema}.${table}.${seq}.csv: ^(?<schema>[^.]+)(\.)(?<table>[^.]+)(\.)([^.]+\.)csv$

OBLOADER locates the files to be imported based on the correspondence between the table names and the filenames. For example, if the target table name is t1 and the specified format is --csv. In this case, the default file suffix is .csv. You can use the --file-suffix option to specify a custom file suffix. OBLOADER will traverse the path specified by -f. Files that match the table binding relationship will include t1.csv, t1.0.csv, and t1.1.2.csv.

If the filenames of the files to be imported follow other rules or no naming rules, you can use the --file-regular-expression option to specify a regular expression to match the files to be imported. For example, --file-regular-expression=".*\.csv" matches all filenames that end with .csv. --file-regular-expression=".*" matches all files in the specified directory.

Option introduction

Option style

OBLOADER provides parameters in the Unix and GNU styles.

• In the Unix style, a single hyphen (-) is used before the parameter, and the option is a single character. For example, ps -e. In this style, you can omit the space between the option and parameter. For example, -p******.

• In the GNU style, a double hyphen (--) is used before the parameter, and the option is a single character or a string. For example, ps --version. In this style, a space must be placed between the option and parameter. For example, --table 'test'.

Option categories

OBLOADER options are classified into basic options and advanced options.

• Basic options: These are commonly used OBLOADER options, including connection options, feature options, and other options.

• Advanced options: These include feature options, performance options, and other options.

Required options

To use OBLOADER to import data, you must specify the connection options, format options, database object type options, and storage path options.

Example:

$./obloader -h xx.x.x.x -P 2883 -u test@mysql#cluster_a -p ****** -D USERA --csv --table '*' -f /output

Here, -h, -P, -u, -p, and -D are connection options; --csv is a file format option; --table is a database object type option; and -f specifies the storage path.

Basic options

Connection options

OBLOADER needs to connect to OceanBase Database to read and write data. You can specify the following options to connect to OceanBase Database.

Import types

Direct load

OBLOADER V4.2.6 and later support the direct load mode. This feature supports OceanBase Database V4.2.1.7 and later in the V4.2.1.x series, V4.2.4.x to V4.3.0.x, and V4.3.0.1 and later in the V4.3.x series.

• --rpc-port= rpc_port_num

  Specifies the RPC port of the OBServer node. This option is used in conjunction with --direct and --parallel to connect to the RPC port of the OBServer node for data import in the direct load mode.

  Note

  • This option applies only to OceanBase Database V4.2.0 RC2 and later.
  • You can obtain the OBServer RPC port in the following ways:
    • When connecting to ODP:
      • In the cloud-based OceanBase environment, the default ODP RPC port is 3307.
      • In the OceanBase Database environment, the default port is 2885. If you want to specify a custom port, you can use the -s option when starting ODP.
    • When connecting to OBServer, you can query the DBA_OB_SERVERS view in the sys tenant to obtain the RPC port of OBServer. The default port is 2882.

  • --direct

    Enables the direct load mode. This option is used in conjunction with --rpc-port and --parallel.

    --direct uses different direct load modes depending on whether the target table contains data.

    • When the target table is empty, --direct uses the full direct load mode. It can also be used with --replace-data, which specifies that data conflicts should be resolved by replace.

    • When the target table is not empty, --direct uses the incremental direct load mode. If the target table has no indexes or LOB fields, --replace-data can be used. If the target table has indexes or LOB fields, --replace-data cannot be used.

    Note

    • The binary data types are not supported in the OBLOADER direct load mode.
    • The OBLOADER direct load mode supports connecting to OBServer and ODP. The versions must meet the following requirements:
      • When connecting to OBServer, OBServer must be V4.2.1.7 or later of the V4.2.1.x series, or between V4.2.4.0 and V4.3.0, or V4.3.0.1 or later of the V4.3.x series.
      • When connecting to ODP, ODP must be V4.3.0 or later, and OBServer must be V4.2.1 or later.

  • --parallel= parallel_num

    Specifies the degree of parallelism for loading data in the direct load mode. This option is used in conjunction with --rpc-port and --direct.

  Feature options

  File formats

  Format (command-line option)	Related command-line options	Scenarios
  --csv Specifies to import a CSV file.	--skip-header Specifies whether to skip the header row of a CSV file. The value is BOOLEAN. --column-delimiter 'column_delimiter_char' Specifies the string delimiter. --line-separator 'line_separator_string' Specifies the line separator. --escape-character 'escape_char' Specifies the escape character. --column-separator 'column_separator_char' Specifies the column separator in the CSV format. The default value is a comma (,). --with-trim Deletes the left- and right-side spaces. --ignore-unhex Indicates whether to ignore the decoding of hexadecimal strings. This option applies only to binary data types. --character-set 'character_set_string' Specifies the character set for creating a database connection. --null-string 'null_replacer_string' Replaces a specified string with NULL. --empty-string 'empty_replacer_string' Replaces a specified string with an empty string (' '). The default value is \E. --file-suffix 'suffix_name' Specifies the file suffix name when importing data. --file-encoding 'encode_name' Specifies the file encoding used to read the data file, which is not the same as the database encoding.	Applicable to almost all business scenarios.
  --cut Specifies to import a CUT file.	--skip-header Specifies whether to skip the header row of a CUT file. The value is BOOLEAN. --skip-footer Specifies whether to skip the last row of a CUT file. --column-splitter 'split_string' Specifies the column separator in the CUT format. --trail-delimiter Indicates whether to add a delimiter at the end of each data row. --line-separator 'line_separator_string' Specifies the line separator. --with-trim Deletes the left- and right-side spaces. --ignore-escape Specifies whether to ignore the escape characters when importing a CUT file. The default value is false. --ignore-unhex Indicates whether to ignore the decoding of hexadecimal strings. This option applies only to binary data types. --character-set 'character_set_string' Specifies the character set for creating a database connection. --null-string 'null_replacer_string' Replaces a specified string with NULL. --escape-character 'escape_char' Specifies the escape character. --empty-string 'empty_replacer_string' Replaces a specified string with an empty string (' '). The default value is \E. --file-suffix 'suffix_name' Specifies the file suffix name when importing data. --file-encoding 'encode_name' Specifies the file encoding used to read the data file, which is not the same as the database encoding.	Similar to CSV. Often used to import flexible data exchange formats exported by some older platforms.
  --pos Specifies to import a POS file.	Must be used with a control file.	For interactions with data unload platforms that only support fixed-length formats.
  --sql Specifies to import an SQL file.	--line-separator 'line_separator_string' Specifies the line separator. --ignore-unhex Indicates whether to ignore the decoding of hexadecimal strings. This option applies only to binary data types. --character-set 'character_set_string' Specifies the character set for creating a database connection. --null-string 'null_replacer_string' Replaces a specified string with NULL. --file-suffix 'suffix_name' Specifies the file suffix name when importing data. --file-encoding 'encode_name' Specifies the file encoding used to read the data file, which is not the same as the database encoding.	Applicable to scenarios where only data is imported, not schemas. Supports only one SQL syntax but offers better performance.
  --mix Specifies to import a MIX file.	--character-set 'character_set_string' Specifies the character set for creating a database connection. --file-suffix 'suffix_name' Specifies the file suffix name when importing data. --file-encoding 'encode_name' Specifies the file encoding used to read the data file, which is not the same as the database encoding. --compat-mode Performs compatible table schema definition imports for MySQL.	Applicable to SQL files of any formats. Offers better compatibility but significantly lower performance compared to --sql. The current implementation reads all text into memory before parsing, making it unsuitable for handling large files.
  --par Specifies to import a Parquet file.	--character-set 'character_set_string' Specifies the character set for creating a database connection. --ignore-unhex Indicates whether to ignore the decoding of hexadecimal strings. This option applies only to binary data types. --file-suffix 'suffix_name' Specifies the file suffix name when importing data. --file-encoding 'encode_name' Specifies the file encoding used to read the data file, which is not the same as the database encoding.	Data migration from Hive, Spark, or other standard Parquet data sources.
  --orc Specifies to import an ORC file.	--character-set 'character_set_string' Specifies the character set for creating a database connection. --ignore-unhex Indicates whether to ignore the decoding of hexadecimal strings. This option applies only to binary data types. --file-suffix 'suffix_name' Specifies the file suffix name when importing data. --file-encoding 'encode_name' Specifies the file encoding used to read the data file, which is not the same as the database encoding.	Data migration from Hive, Spark, or other standard ORC data sources.
  --avro Specifies to import an Avro file.	--file-suffix 'suffix_name' Specifies the file suffix name when importing data.	Often used in big data scenarios, supporting AP business offline migration and other scenarios.

  Note

  For more information about command-line options, see the Options section in this topic.

  Import a file compressed and exported by using OBDUMPER

  OBLOADER V4.3.0 and later support the import of files compressed and exported by using OBDUMPER. Supported importable file formats include CSV, CUT, POS, and SQL. You can configure related settings through the following command-line options.

  • --compress

    A boolean value that indicates whether the file is a compressed file exported by using OBDUMPER.

  • --compression-algo

    A string that indicates the compression algorithm used when a file is exported by using OBDUMPER. Valid values: zstd, zlib, gzip, and snappy. Default value: zstd.

  • --compression-level

    An integer that indicates the compression level used when a file is exported by using OBDUMPER. The supported compression levels vary for different compression formats, as described below:

    • zstd: 1 ~ 22. Default value: 3.

    • zlib: -1 ~ 9. Default value: -1.

    • gzip and snappy: not supported.

  Database object types

  • --all

    Specifies to import all supported database objects and table data. When used with the --ddl option, it specifies to import all database object definitions. When used with any data format option such as --csv, --sql, --cut, or --pos, it specifies to import all data files in the corresponding format. To import all database object definitions and table data, you can use the --all, --ddl, and any data format option.

    Note

    The --all option is mutually exclusive with any database object options. You cannot specify both. If you specify both the --all option and any database object option, the --all option will take precedence.

  • --table-group 'table_group_name [,table_group_name...]' | --table-group '*

    Specifies to import table group definitions. It does not support data import. For other descriptions, refer to the --table option.

  • --table 'table_name [,table_name...]' | --table '*

    Specifies to import table definitions or data. When used with the --ddl option, it specifies to import only table definitions. When used with any data format option, it specifies to import only table data. To import multiple tables, separate the table names with commas (,). By default, table names imported to an Oracle compatible tenant of OceanBase Database are in uppercase, and those imported to a MySQL compatible tenant of OceanBase Database are in lowercase. For example, in the Oracle compatible mode of OceanBase Database, --table 'test' and --table 'TEST' both refer to the TEST table. In the MySQL compatible mode of OceanBase Database, --table 'test' and --table 'TEST' both refer to the test table. If you want to distinguish between uppercase and lowercase, enclose the table name in square brackets ([ ]). For example, --table '[test]' refers to the test table, and --table '[TEST]' refers to the TEST table. If a star (*) is specified, it specifies to import all table definitions or data.

    Note

    When a control file is used, if the table name specified by the --table option is in a different case from that in the database, the control file will not take effect.

  • --view 'view_name [, view_name...]' | --view '*

    Specifies to import view definitions. It does not support data import. For other descriptions, refer to the --table option.

  • --trigger 'trigger_name [, trigger_name...]' | --trigger '*

    Specifies to import trigger definitions. It does not support data import. For other descriptions, refer to the --table option.

  • --obj-user 'user_name [, user_name...]' | --obj-user '*'

    Specifies to import user definitions. It does not support data import. For other descriptions, refer to the --table option.

    • User requirement: The login user must be an administrator or a user with the dba_users privilege in the Oracle compatible mode of OceanBase Database, or an administrator or a user with the mysql.user table privilege in the MySQL compatible mode of OceanBase Database.

    • By default, the following users are ignored: ROOT, SYS, ORAAUDITOR, LBACSYS, proxyro, and standbyro.

  • --role 'role_name [, role_name...]' | --role '*'

    Specifies to import role definitions. It does not support data import. For other descriptions, refer to the --table option.

    • User requirement: At present, it applies only to the Oracle compatible mode of OceanBase Database. The login user must be an administrator or a user with the dba_users privilege.

    • By default, the following roles are ignored: CONNECT, RESOURCE, DBA, PUBLIC, and STANDBY_REPLICATION.

  • --sequence 'sequence_name [, sequence_name...]' | --sequence '*

    Specifies to import sequence definitions. It does not support data import. For other descriptions, refer to the --table option. It is currently applicable only to OceanBase Database in Oracle compatible mode.

  • --synonym 'synonym_name [, synonym_name...]' | --synonym '*

    Specifies to import synonym definitions. It does not support data import. For other descriptions, refer to the --table option. It is currently applicable only to OceanBase Database in Oracle compatible mode.

  • --type 'type_name [, type_name...]' | --type '*

    Specifies to import type definitions. It does not support data import. For other descriptions, refer to the --table option. It is currently applicable only to OceanBase Database in Oracle compatible mode.

  • --type-body 'typebody_name [, typebody_name...]' | --type-body '*

    Specifies to import type body definitions. It does not support data import. For other descriptions, refer to the --table option. It is currently applicable only to OceanBase Database in Oracle compatible mode.

  • --package 'package_name [, package_name...]' | --package '*

    Specifies to import package definitions. It does not support data import. For other descriptions, refer to the --table option. It is currently applicable only to OceanBase Database in Oracle compatible mode.

  • --package-body 'packagebody_name [, packagebody_name...]' | --package-body '*

    Specifies to import package body definitions. It does not support data import. For other descriptions, refer to the --table option. It is currently applicable only to OceanBase Database in Oracle compatible mode.

  • --function 'function_name [, function_name...]' | --function '*

    Specifies to import function definitions. It does not support data import. For other descriptions, refer to the --table option.

  • --procedure 'procedure_name [, procedure_name...]' | --procedure '*

    Specifies to import stored procedure definitions. It does not support data import. For other descriptions, refer to the --table option.

  • --ddl
    Specifies to import DDL files. DDL files can be imported in parallel in OceanBase Database V4.2.1 and later.

    • --character-set 'character_set_string' specifies the character set for creating a database connection.

    • --ignore-unhex specifies to ignore the decoding of hexadecimal strings. This option applies only to binary data types.

    • --file-suffix 'suffix_name' specifies the file extension for data files during import.

    • --file-encoding 'encode_name' specifies the file encoding used to read data files, which is not the same as the database encoding.

  • --source-type hive

    Specifies to import partition information from the Hive storage path. During import, directory names containing = will be parsed into key=value format, where key represents the column name and value represents the column value. Supported file formats include ORC and Parquet. For an example of how to use this parameter, see Command-line application example.

  • --auto-column-mapping

    Specifies to enable automatic column mapping. In this mode, table data from the source file will be imported to the target table column by column. The number of columns in the source file can be different from that in the target table. Supported file formats include CSV, ORC, and Parquet. For an example of how to use this parameter, see Command-line application example.

    Note

    You cannot specify both the --ctl-path and --auto-column-mapping options.

  Storage path

  • -f 'file_path', --file-path= 'file_path'

    Specifies the absolute path for storing data files. If the files are stored in a non-local file system, you can use --tmp-path to specify a temporary directory.

    OBLOADER allows you to import database object definitions and table data from local disks, Alibaba Cloud Object Storage Service (OSS), Amazon Simple Storage Service (S3), Apache Hadoop, Tencent Cloud Object Storage (COS), Huawei Cloud Object Storage Service (OBS), Azure Blob Storage, and Google Cloud Storage (GCS). The syntax is as follows:

    [scheme://host]path[?parameters]
    parameters: key[=value]&...
    

    Component	Description
    scheme	The storage scheme. Supported storage schemes are Alibaba Cloud OSS, Amazon S3, Apache Hadoop, COS, and OBS. If the specified scheme is not one that is listed above, an error is returned.
    host	The name of the storage space. When you import data from OSS, S3, COS, or OBS, the host parameter specifies the bucket. For more information, see OSS Bucket. When you import data from Apache Hadoop, the host parameter specifies the Hadoop node, which is in the <ip>:<port> or <cluster_name> format.
    path	The path for importing data from the storage space. The path must start with a /.
    parameters	The parameters required for the request. The parameters can be a single key or a key-value pair.

    The following table describes the parameters.

    Parameter	Required	Description	Supported storage type	Supported from
    endpoint	Yes	The endpoint of the region where the host is located. You can access OSS, S3, COS, and OBS services via the domain name endpoint. Example: oss://ssmp-xxxx-xxxx/test?endpoint=oss-cn-shenzhen-internal.aliyuncs.com.	OSS/S3/COS/OBS	OSS: V4.2.0 S3: V4.2.5 COS and OBS: V4.3.0
    region	Yes	The endpoint specifies the physical location of the bucket.	OSS, S3, COS, and OBS	OSS and S3: V4.2.0 COS and OBS: V4.3.0
    storage-class	Yes	The Amazon S3 storage class.	S3	V4.2.0
    access-key	Yes	The access account for the storage.	OSS, S3, COS, and OBS	OSS and S3: V4.2.0 COS and OBS: V4.3.0
    secret-key	Yes	The access key for the storage.	OSS, S3, COS, and OBS	OSS and S3: V4.2.0 COS and OBS: V4.3.0
    hdfs-site-file	Yes	The configuration file that contains the configuration information of Apache Hadoop, such as the block size and the number of replicas.	Hadoop	V4.2.1
    core-site-file	Yes	The configuration file that contains the core configuration information of the Hadoop cluster, such as the URI of the Apache Hadoop file system and the default file system.	Hadoop	V4.2.1
    principal	Yes	The authentication identifier in Kerberos.	Hadoop	V4.2.1
    keytab-file	Yes	The absolute path of the Keytab file, which is used to authorize users or services to access the system resources.	Hadoop	V4.2.1
    krb5-conf-file	Yes	The path of the Kerberos configuration file.	Hadoop	V4.2.1

    Here are some examples:

    • Import data from a local disk.

      -f '/home/admin/foo/bar' 
      

    • Import data from Amazon S3.

      -f 's3://bucket/path?region={region}&access-key={accessKey}&secret-key={secretKey}'
      

    • Import data from Alibaba Cloud OSS.

      -f 'oss://mybucket/foo/bar?endpoint=myendpoint&access-key=myak&secret-key=mysk' 
      

    • Import data from Hadoop.

      -f 'hdfs://***.*.*.*:9000/chang/parquet?hdfs-site-file=/data/0/zeyang/hdfs-site.xml&core-site-file=/data/0/zeyang/core-site.xml
      

    • Import data from Tencent Cloud Object Storage.

      -f 'cos://mybucket/foo?region=ap-shanghai&access-key=myak&secret-key=mysk'
      -f 'cos://mybucket/foo?endpoint=yourendpoint&access-key=myak&secret-key=mysk'
      

    • Import data from Huawei Cloud Object Storage Service.

      -f 'obs://mybucket/foo?region=cn-north-1&access-key=myak&secret-key=mysk'
      -f 'obs://mybucket/foo?endpoint=yourendpoint&access-key=myak&secret-key=mysk'
      

    • Import data from Azure Blob Storage.

      Azure Blob Storage supports the wasb:// and wasbs:// (SSL encrypted access) protocols. You can import data in the following ways.

      • Append the sas token to the URI of the wasb protocol as the value of --file-path option.

        wasb[s]://{container}@{accountName}.{endpointSuffix}/path/file ? SAS
        

        Here is an example:

        wasbs://test-data@loader.blob.core.chinxxxpi.cn/linxxx.csv?sv=2022-11-02&ss=bfqt&srt=sco&sp=rwdlacupitfx&se=2024-08-13T14:49:32Z&st=2024-08-13T06:49:32Z&spr=https&sig=m2AXglpESQPYrVvrhxxxx
        

      • Append the connection string to the URI of the wasb protocol as the value of --file-path option. Except for account key, other parameters are optional.

        wasb[s]://{container}@{accountName}.{endpointSuffix}/path/file ? account-key=xxxx
        
        wasb[s]://{container}@{accountName}.{endpointSuffix}/path/file ? default-endpoints-protocol=xxx & account-key=xxxx & account-name=xxx & endpoint-suffix=xxx
        

        Here is an example:

        wasbs://test-data@loader.blob.core.chixxxx.cn/lineitem1.csv?account-key=xxxx
        

    • Import data from Google Cloud Storage.

      -f 's3://bucket/path?region={region}&access-key={accessKey}&secret-key={secretKey}'
      

  • --ctl-path 'control_path'

    Specifies the absolute path for storing the control file on the local disk. The control file stores the built-in processing functions. The data is preprocessed based on the configured functions before the data is imported. For example, the preprocessing includes case conversion and null value check. For more information about the control file, see the Data processing section. This option requires that you specify the value in single quotation marks. For example: --ctl-path '/home/controls/'.

    Notice

    You cannot use the --ctl-path and --auto-column-mapping options at the same time.

  • --log-path 'log_path'

    Specifies the output directory for OBLOADER operation logs. If you do not specify this option, OBLOADER operation logs are output to the directory specified by the -f option. In most cases, you do not need to redirect the logs.

  Other options

  • -H, --help

    View the help message of the CLI tool.

  • -V, --version

    View the version number of the current tool.

  Advanced options

  Feature options

  Timestamp formats

  • --nls-date-format 'date-format-string'

    Specifies the date format for a database connection in the Oracle compatible mode of OceanBase Database. Default value: YYYY-MM-DD HH24:MI:SS.

  • --nls-timestamp-format 'timestamp-format-string'

    Specifies the timestamp format for a database connection in the Oracle compatible mode of OceanBase Database. Default value: YYYY-MM-DD HH24:MI:SS:FF9.

  • --nls-timestamp-tz-format 'timestamp-tz-format-string'

    Specifies the timestamp with time zone format for a database connection in the Oracle compatible mode of OceanBase Database. Default value: YYYY-MM-DD HH24:MI:SS:FF9 TZR.

  Allowlist and blocklist filtering

  • Table allowlists and blocklists

    • --exclude-table 'table_name [, table_name...]'

      Specifies the table names to be excluded when you import table definitions or data. Table names can be specified by using wildcard characters.

      Example: --exclude-table 'test1,test*,*test,te*st'

      In this example, the following tables are excluded when you import table definitions or data:

      • test1

      • All tables whose names start with test

      • All tables whose names end with test

      • All tables whose names start with te and end with st

    • --exclude-data-types 'datatype [, datatype...]'

      Specifies the data types to be excluded when you import data.

    • --file-regular-expression

      Allows you to specify a regular expression of file names to selectively import files. Example: --file-regular-expression ".*" matches all files in the directory by using the ".*" regular expression, where * matches zero or more arbitrary characters. Example: --file-regular-expression ".*\\.csv" matches files whose names end with .csv by using the ".*\\.csv" regular expression, where "\." matches the actual period character ".".

  • Column allowlists and blocklists

    • --include-column-names 'column_name [, column_name...]'

      Allows you to specify the column names for data import.

    • --exclude-column-names 'column_name [, column_name...]'

      Allows you to specify the column names for data exclusion.

      Notice

      • The specified column names must be in the same case as the column names in the table schema.
      • The number of data columns in the data file must be the same as the number of remaining columns after the allowlist or blocklist is applied, and they must be in the same order. If the data file is in the CSV, ORC, or Parquet format, we recommend that you use it with the --auto-column-mapping option.
      • The control file cannot be used together with the --exclude-column-names option. The functionality of the --exclude-column-names option is included in the control file.

  Error handling

  • --max-discards int_num

    Specifies the maximum number of duplicates allowed for each table. If the number of duplicates in any table exceeds this limit, the data import for that table is stopped, and the log records the failure of importing that table. This does not affect the data import of other tables. The default value is -1, which indicates to continue importing data regardless of duplicates.

    Notice

    • This option takes effect only when a table has a primary key or unique key and contains duplicate data. It only applies to errors that occur when writing data to the database, not to other types of errors (such as errors that occur when reading data from the database file).
    • This option is not applicable to direct load.

  • --retry

    Specifies to resume data import from the interrupted point. We recommend that you use this option only when more than 80% of the data has been imported. In this way, you can avoid re-importing the data and reduce the risk of importing a few pieces of duplicate data. Otherwise, you can clear the table and start data import again. This method is more efficient.

    Note

    • The load.ckpt file is a savepoint file generated by the tool during its operation. The path of this file is located in the directory specified by -f. If the load.ckpt file does not exist, this option cannot be used.
    • This option is not applicable to direct load.

  • --max-errors int_num

    Specifies the maximum number of errors allowed for each table during data import. If the number of errors in importing data of any table exceeds this limit, the data import for that table is stopped, and the log records the failure of importing that table. Valid values of this option are 0, -1, and a positive integer N. If you set this option to -1, errors are ignored and data import continues. The default value is 1000.

    Notice

    • This option only applies to errors that occur when writing data to the database, not to other types of errors (such as errors that occur when reading data from the database file).
    • This option is not applicable to direct load.

  • --strict= 'strict_string'

    Controls the impact of dirty data on the exit status of the process during data import. The default value is true. In this case, if the imported data contains bad records or discarded records, the program exits in failure (System exit 1). If you set the value to false, the exit status of the program is not affected by bad records or discarded records in the imported data. The program exits in success (System exit 0).

    Note

    You can use this option in combination with the --max-discards or --max-errors option. In this way, when the number of duplicates or errors is within the specified range, the system will skip the errors and continue the process. For more information, see Error handling.

  • --replace-data

    Specifies to replace the duplicate data in the table. This option applies only to tables with defined primary keys or unique keys (including non-null fields). If the file contains a large amount of duplicate data, which exceeds 30% of the total data, we recommend that you clear the table and start data import again. Data replacement is less efficient than data import into an empty table. This option takes effect only when it is used together with the --csv, --sql, --cut, --par, or --orc option, but not with the --ddl option.

    Note

    • If the file and the table contain duplicate data, the data in the table will be replaced with that in the file.
    • If a table does not have a primary key or unique key, this option will append data to the table.
    • Do not specify this option in the command line if you do not want to replace the duplicate data, to avoid impact on your business.

  Performance options

  • --rw float_num

    Specifies the ratio of the number of file parsing threads to the total number of threads. Default value: 1.

    This option is used in combination with the --thread option to calculate the number of file parsing threads as the value of --thread option multiplied by the value of --rw option.

  • --slow float_num

    Specifies the threshold for the tool to enter the slow load mode. When the memory usage of OceanBase Database reaches 75%, OBLOADER automatically reduces the speed to prevent the rapid increase of database memory usage. Default value: 0.75.

  • --pause float_num

    Specifies the threshold for the tool to enter the pause load mode. When the memory usage of OceanBase Database reaches 85%, OBLOADER automatically stops the import to prevent high database memory water level from causing memory issues. Default value: 0.85.

  • --batch int_num

    Specifies the transaction size for batch writing. We recommend that you set the value of this option to a smaller one in proportion to the width of the table. The value of this option must not cause an overflow of database memory. Default value: 200.

    Note

    • OBLOADER V4.2.0 and later versions support adaptive --batch default values based on the memory of the Java Virtual Machine (JVM).

    • If you use OBLOADER V4.2.8.1 or V4.2.8.2, the actual number of parameters in the prepared statement may exceed the limit, which can cause an ODP core. If this issue occurs, you can manually set this option to a reasonable value.

  • --thread int_num

    Specifies the number of concurrent threads. This option directly corresponds to the number of write threads. When used in combination with the --rw option, it is used to calculate the number of file parsing threads, calculated as the value of the --thread option multiplied by the value of the --rw option. The default value is 2 × CPU. If the CPU is greater than 16, the default maximum value is 32.

    Note

    OceanBase Database V4.2.1 and later versions support parallel import of DDL. When you import database objects defined in versions earlier than OceanBase Database V4.2.1, you do not need to specify this option.

  • --block-size

    Specifies the threshold for splitting file blocks, which supports the LONG data type. For example, --block-size 64 indicates that the size of a single file does not exceed 64 MB. When you specify this option, you do not need to explicitly indicate the unit. The default unit is MB. The default value is 64. By default, OBLOADER automatically splits large files into multiple logical subfiles based on 64 MB without occupying additional storage space.

  • --max-tps int_num

    Specifies the maximum TPS. When you specify this option, it can ensure a stable import rate.

  • --max-wait-timeout int_num

    Specifies the maximum waiting time for OceanBase Database to perform major compactions. When you specify this option, you do not need to explicitly indicate the unit. The default unit is hours. When OceanBase Database is in the major compaction mode, OBLOADER is in the pause load mode. The maximum waiting time specified by this option cannot be exceeded. The default value is 3.

  • --mem

    Supported in OBLOADER V4.3.2 and later versions, this option specifies the memory size for the JVM. The supported units are K, M, G, and T. The default value is 4G.

  Other options

  • --truncate-table

    Specifies to truncate tables in the target database that contain data files before importing data. This option can be used only with one of the data format options. If it is used with --all or --table '*', it specifies to truncate all tables that contain data files in the target database. You can explicitly specify --table 'test1,test2,...' to truncate only specific tables.

    If you use this option, the system will ask you to confirm the operation. You must enter Y to confirm.

    Notice

    Do not use this option to truncate the target database or tables in it. We recommend that you manually truncate tables based on your business needs to avoid impacting your business.

  • --delete-from-table

    Specifies to delete tables in the target database that contain data files before importing data. This option can be used only with one of the data format options. If it is used with --all or --table '*', it specifies to delete all tables that contain data files in the target database. You can explicitly specify --table 'test1,test2,...' to delete only specific tables.

    If you use this option, the system will ask you to confirm the operation. You must enter Y to confirm.

    Notice

    • If you use the --all or --table '*' option together with the --delete-from-table option, the program will delete all tables that contain data in the target database, regardless of whether the directories specified by the -f option contain data files of the tables.
    • Do not use this option to delete data in the target database or tables. Especially do not use it to delete data in large tables. We recommend that you manually delete data based on your business needs to avoid impacting your business.

  • --yes

    When you use the --truncate-table or --delete-from-table option, the system will ask you to confirm the operation to remind you of the risk. If you do not want to confirm interactively, you can use the --yes option to ignore the confirmation, so that the system will not prompt you to confirm the risk.

  • --replace-object

    Specifies to replace existing database object definitions with those defined in the files when you import database objects. Objects such as tables and synonyms are deleted and then created, and objects such as functions and stored procedures are replaced by using the CREATE OR REPLACE statement. This option can be used only with the --ddl or --mix option; it does not take effect with other data format options such as --csv or --sql.

    Notice

    • Objects in the target database will be forcibly replaced with those defined in the files.
    • Do not use this option if you do not want to replace database objects, to avoid impacting your business.

  • --session-config

    Specifies the connection configuration file. The configuration file directory provides a default configuration file: <root directory of the tool>/conf/session.config.json. The file takes effect without configuration. We recommend that you specify this option only when you need to load multiple connection configurations into the same configuration file directory.

  • --default-date

    The alternative value for invalid dates. No default value is provided. The date value is formatted based on the --date-format option. If the formatting fails, the date alternative value is returned.

    Notice

    This parameter takes effect only on the DATE and DATETIME data types in MySQL compatible mode of OceanBase Database.

  • --date-format

    The datetime format of the original data. The default value is yyyy-MM-dd HH:mm:ss. It is used to verify whether the data in the file is valid date time data. If the datetime data is invalid, the value specified by --default date is used as the default datetime.

    Notice

    This parameter takes effect only on the DATE and DATETIME data types in MySQL compatible mode of OceanBase Database.

  Options

  Option	Required	Description	Introduced in	Deprecated in
  -h(--host)	Yes	The IP address of the host to connect to ODP or an OceanBase physical node. If the IP address is an IPv6 address, enclose it in square brackets ([]). Example: [2001:0db8:85a3:xxxx:xxxx:8a2e:0370:7334].
  -P(--port)	Yes	The port to connect to ODP or an OceanBase physical node.
  --rpc-port	No	The port for connecting to the OBServer node's RPC service.	V4.2.5
  --compat-mode	No	Imports the table schema defined in MySQL compatible mode of OceanBase Database. When importing a table schema defined in MySQL, OBLOADER converts the native MySQL table creation statement to an equivalent statement supported by the MySQL compatible mode of OceanBase Database. Notice Only MySQL table schemas can be imported. Only MySQL Database V5.6, V5.7, and V8.0 are supported.	V4.2.7
  --direct	No	Specifies the direct load mode.	V4.2.5
  --parallel	No	The degree of parallelism of data loading in direct load mode.	V4.2.6
  -u(--user)	Yes	The username, tenant name, and cluster name. Format: <user>@<tenant>#<cluster>.
  -p(--password)	No	The database password.
  -D(--database)	Yes	The database name.
  -f(--file-path)	Yes	The directory where the data file is stored or the absolute path of the data file.
  --sys-user	No	The username under the sys tenant.
  --sys-password	No	The password of a specific user in the sys tenant.
  -t	No	The tenant name for connecting to OceanBase Database. When you perform direct load in the cloud, you must use the --public-cloud -t option. Notice When performing direct load in the cloud, you must use the --public-cloud -t option.
  --public-cloud	No	The option for importing database objects or table data from an OceanBase cluster deployed in the cloud. Notice When performing direct load in the cloud, you must use the --public-cloud -t option.
  --file-suffix	No	The file extension name. Generally, the file extension name is associated with the file format. For example, the extension name of a CSV file is usually .csv. If you do not strictly follow this naming convention, you can name a CSV file with any extension, such as .txt. In this case, OBLOADER cannot identify the target file based on the format. The --file-suffix option is optional, and each data format has a corresponding default value. By default, the file extension name for CSV format is .csv, for SQL format is .sql, for CUT format is .dat, and for POS format is .dat. When specified on the command line, enclose the value in single quotation marks. Example: --file-suffix '.txt'.
  --file-encoding	No	The file encoding (different from the database encoding). When specified on the command line, enclose the value in single quotation marks. Example: --file-encoding 'GBK'. Default value: UTF-8.
  --ctl-path	No	The directory where the control file is stored.
  --log-path	No	The output directory for log files.
  --ddl	No	The option for importing DDL files. Starting from OceanBase Database V4.2.1, OBLOADER supports parallel import of DDL files. A DDL file is a database object definition file, and its naming convention is object name-schema.sql. When specified on the command line, OBLOADER imports only the database object definitions and does not import any table data. Notice Do not include comments or switches in the file. If the database objects are dependent on each other, the import may fail and manual intervention will be required.
  --csv	No	The option for importing CSV files. (Recommended) A CSV file is a data file stored in the standard CSV format, and its naming convention is table name.csv. For more information about the CSV format, see the definition in RFC 4180. The most common error in the CSV format is incorrect delimiters. Single or double quotation marks are commonly used as delimiters. If the data contains delimiters, escape characters must be specified. Otherwise, OBLOADER cannot parse the data. We recommend that you use the CSV format. We recommend that you use the --csv option in combination with the --table option. If you use the --all option in combination with it, OBLOADER will import only the data file corresponding to the table and will not import any database object definitions.
  --sql	No	The option for importing SQL files. (Different from DDL files) An SQL file is a data file stored in the format of an INSERT statement, and its naming convention is table name.sql. The content of the file consists of executable INSERT statements for each row of table data. OBLOADER V4.3.1 and later support the import of SQL text files in the batch insert format. The content format of an SQL file is clearly different from that of a DDL file. We recommend that you use the --sql option in combination with the --table option. If you use the --all option in combination with it, OBLOADER will import only the data file corresponding to the table and will not import any database object definitions. Notice The data must not contain SQL functions, special characters, or line breaks. Otherwise, the file may not be properly parsed.
  --orc	No	The option for importing ORC files. An ORC file is a data file stored in columnar format, and its naming convention is table name.orc. Please refer to the Apache ORC format definition.	V4.0.0
  --par	No	The option for importing Parquet files. A Parquet file is a data file stored in columnar format, and its naming convention is table name.parquet. Please refer to the Apache Parquet format definition. Note When using OBLOADER V4.2.5 or earlier to import Parquet files, the DECIMAL, DATE, TIME, and TIMESTAMP data types are not supported.	V4.0.0
  --mix	No	The option for importing MIX files. A MIX file is a file that contains both DDL and DML statements. The file does not follow a strict naming convention. Notice MIX files do not follow a strict format, the processing is complex, and the performance is poor. We recommend that you do not use them.
  --pos	No	The option for importing POS files. A POS file is a data file stored in fixed-length byte format, and its naming convention is table name.dat. Fixed-length bytes mean that each column of data is stored using a fixed number of bytes. Extra spaces are added if the data is shorter than the fixed length, and the excess data is truncated if it is longer than the fixed length. Truncating data by byte length may result in data corruption. We recommend that you use the --pos option in combination with the --table option. If you use the --all option in combination with it, OBLOADER will import only the data file corresponding to the table and will not import any database object definitions. (This is different from fixed-length character format).
  --cut	No	The option for importing CUT files. A CUT file is a data file stored in a format where columns are separated by single characters or strings, and its naming convention is table name.dat. How do you identify a CUT file from a CSV file? In a CSV file, fields are separated by single characters, and the commonly used separator is a comma. In a CUT file, fields are separated by strings. For example, the fields are separated by the string |@|. In a CSV file, single or double quotation marks are used as delimiters between fields. A CUT file does not have delimiters. We recommend that you use the --cut option in combination with the --table option. If you use the --all option in combination with it, OBLOADER will import only the data file corresponding to the table and will not import any database object definitions. Notice Data in a CUT file must be stored in whole lines. If the field separator is a single character, avoid using the separator, carriage return, or line break in the data. Otherwise, OBLOADER cannot correctly parse the data.
  --avro	No	The option for importing Avro files. An Avro file is an Avro file that conforms to the Apache standard and has the extension name .avro. The data types supported for import and the limitations are as follows: All primary types are supported. All logical types are supported. Complex types are not supported.	V4.3.0
  --all	No	The option for importing all supported database objects and table data.
  --table-group	No	The option for importing table group definitions.	V3.1.0
  --table	No	The option for importing table definitions or table data.
  --view	No	The option for importing view definitions.
  --trigger	No	The option for importing trigger definitions.
  --obj-user	No	The option for importing user definitions.
  --role	No	The option for importing role definitions.
  --sequence	No	The option for importing sequence definitions.
  --synonym	No	The option for importing synonym definitions.
  --type	No	The option for importing type definitions.	V4.0.0
  --type-body	No	The option for importing type body definitions.
  --package	No	The option for importing package definitions.
  --package-body	No	The option for importing package body definitions.
  --function	No	The option for importing function definitions.
  --procedure	No	The option for importing stored procedure definitions.
  --replace-object	No	The option for replacing existing object definitions. (Not recommended. We recommend that you manually replace them.)
  --rw	No	The ratio of parsing threads for data files.
  --slow	No	The threshold for triggering a slow import.
  --pause	No	The threshold for triggering a stop import.
  --batch	No	The number of records in each batch transaction.
  --thread	No	The number of concurrent threads in the import task.
  --block-size	No	The size of each file split.
  --retry	No	The option for importing from the last save point.
  --max-tps	No	The import rate limit. Default unit: rows per second.
  --max-wait-timeout	No	The maximum timeout for waiting for database compaction.
  --nls-date-format	No	The session-level date and time format. (Supported only in the Oracle compatible mode of OceanBase Database)
  --nls-timestamp-format	No	The session-level timestamp format. (Supported only in the Oracle compatible mode of OceanBase Database)
  --nls-timestamp-tz-format	No	The session-level timestamp with time zone format. (Supported only in the Oracle compatible mode of OceanBase Database)
  --trail-delimiter	No	The option for deleting the last column separator at the end of a line.
  --with-trim	No	The option for deleting leading and trailing spaces.
  --skip-header	No	A boolean value indicating whether to skip the first line of CSV/CUT files. Only from V3.3.0, the first line of CUT files can be skipped.
  --skip-footer	No	The option for skipping the last line of CUT files.	V3.3.0
  --null-string	No	The string to replace NULL. Default value: \N .
  --empty-string	No	The string to replace empty strings (' '). Default value: \E .
  --line-separator	No	The line separator. When you import CSV, CUT, POS, or SQL files, you can specify the line separator in the data file. The default value of this option depends on the system platform and includes only three possible values: \r, \n, and \r\n.
  --column-separator	No	The column separator in CSV files, which is different from the column separator in CUT files.
  --escape-character	No	The escape character. This option supports only single characters and can be used only with the --csv or --cut option. When used with the --csv option, the default value is null. When used with the --cut option, the default value is \. Note If you specify this option for a special (invisible) character, use its hexadecimal representation. For example, you can use '\x09' to represent an invisible tab character.
  --column-delimiter	No	The string delimiter. This option supports only single characters and can be used only with the --csv option. Default value: single quotation mark('.').
  --ignore-unhex	No	The option for ignoring the decoding of hexadecimal strings.
  --exclude-table	No	The option for excluding table definitions and data.
  --exclude-data-types	No	The option for skipping the import of data of specified data types.
  --column-splitter	No	The column separator string in CUT files, which is different from the column separator in CSV files.
  --max-discards	No	The maximum number of duplicate records allowed for a single table during import. Default value: -1. Notice This option takes effect only when a table has a primary key or unique key and contains duplicate data. It only applies to errors that occur when writing data to the database, not to other types of errors (such as errors that occur when reading data from the database file). This option is not applicable to direct load.
  --max-errors	No	The maximum number of errors allowed for a single table during import. Default value: 1000. Notice This option only applies to errors that occur when writing data to the database, not to other types of errors (such as errors that occur when reading data from the database file). This option is not applicable to direct load.
  --include-column-names	No	The option for importing data based on specified column names.
  --exclude-column-names	No	The option for excluding data corresponding to specified column names during import.
  --replace-data	No	The option for replacing duplicate data. (Applicable only to tables with a primary key or unique key in a non-null column)
  --truncate-table	No	The option for truncating the target table before the import. (Not recommended. We recommend that you manually truncate the table.)
  --delete-from-table	No	The option for deleting all data in the target table before the import. (Not recommended. We recommend that you manually delete the data.)
  --yes	No	The option for ignoring the confirmation prompt after you use the --truncate-table or --delete-from-table option.
  -V(--version)	No	The option for viewing the OBLOADER version number.
  --no-sys	No	The option for indicating that the sys tenant password cannot be provided in an OceanBase Database environment.	V3.3.0
  --logical-database	No	The option for importing into a logical database of ODP (Sharding).	V3.3.0
  --file-regular-expression	No	The option for specifying the regular expression of file names to selectively import files.	V3.3.0 Note V4.3.2 and later support the import of files by using regular expressions.
  --ignore-escape	No	The option for ignoring the escape of characters when importing CUT files.	V3.3.0
  --character-set	No	The character set for creating a database connection. Default value: the value of jdbc.url.character.encoding in the session.properties file. The value specified by the --character-set option overrides the value of jdbc.url.character.encoding. This option supports the following character sets: binary, gbk, gb18030, utf16, and utf8mb4.	V4.2.4
  --strict	No	The option for controlling the impact of dirty data on the exit status of the process during import.	V4.2.4
  --session-config	No	The option for specifying the connection configuration file.	V4.2.6
  --default-date	No	The alternative value for invalid dates. The date value is formatted based on the --date-format option. If the formatting fails, the date value specified by --default date is returned.	V4.3.4.1
  --date-format	No	The datetime format of the original data. The default value is yyyy-MM-dd HH:mm:ss. This option is used to verify whether the data in the file is valid datetime data. If the data is invalid datetime data, the value specified by --default date is used as the default datetime.	V4.3.4.1
  -H(--help)	No	The option for viewing the help of OBLOADER command-line tool.