OBLOADER & OBDUMPER V4.3.2

OBLOADER & OBDUMPER V4.3.2 was released in October 2024 to fix some known issues as well as the issue where no error is returned for incomplete export of non-partitioned tables without primary keys.

Version information

• Current version: V4.3.2

• Release date: October 31, 2024

New features

• The --source-type hive option is provided to allow you to import partition information from a Hive storage path.

• The --auto-column-mapping option is provided to support automatic mapping for column names. With this option specified, data is imported from the source file to the target table based on column names. The number of columns in the source file can be different from that in the target table. CSV, Parquet, and ORC files are supported.

• Complex data types of ORC and Parquet files are supported.

• The --mem 4G option is provided to specify the memory size. This way, you do not need to modify the relevant JVM memory parameter.

• You can set the -Denable.parallel.write startup parameter to true to enable parallel writes. This way, exported files are combined into one.

• The parsing efficiency for CSV files is improved by approximately 30%.

• Mainstream object storage services, including Azure Blob Storage, Google Cloud Storage (GCS), and Huawei Cloud Object Storage Service (OBS), are supported.

• The direct load mode is supported. You can use the --direct option to specify to enable full or incremental direct load depending on whether the target table contains data.

• The --file-regular-expression option is extended to allow you to extract database and table names from file names to support multi-table import.

Compatibility changes

When OBLOADER parses CSV files, the escape character changes from '\' to NULL values by default to comply with CSV specifications.

Bug fixes

• Fixed the issue where the retry mechanism specified by the --retry option is unavailable.

• Fixed the issue where an incorrect table is imported when the --table '*' option is specified for single-file import.

• Fixed the issue where the export format of the JSON data type is invalid in Oracle compatible mode.

• Fixed the issue where the file exported by OBDUMPER from a table contains subnames and cannot be imported by OBLOADER.

• Fixed the issue where counter overflow occurs when 4 billion rows are imported from a server in direct load mode.

• Fixed the issue where OBLOADER returns the following error when the data contains characters in special formats: Error: Data too long for column.

• Fixed the issue where datetime types imported by using control files are inconsistent.

• Fixed the issue where OBLOADER returns the following error when it attempts to restore a DDL file: Executing: "drop table `if`;" skipped. Reason: Unknown table 'test.if'.

• Fixed the issue where an error occurs during the import of a standard CSV file that contains strings with double quotation marks (" ") or slashes (/).