BladePipe 1.8.0: Faster binlog parsing, new pipeline support, and more stable performance.
Skip to main content

BladePipe 1.8.0

Release Date: June 30, 2026

Version: 1.8.0

Highlights​

  • Optimized incremental data parsing for MySQL, PolarDB-X, PolarDB for MySQL, and OceanBase for MySQL sources in binlog mode. This significantly reduces CPU usage when parsing wide tables on the source side.
  • Optimized the incremental DataJob data format for MySQL, PolarDB for MySQL, Oracle, PostgreSQL, and SQL Server sources to reduce DataJob memory usage. New parameters: updateImageMode and deleteImageMode.

New Pipeline​

  • Added support for PolarDB for MySQL > OceanBase for MySQL, including schema migration, full data load, incremental sync, data verification, data correction, and DDL sync (add column / drop column / modify / rename / truncate).

New Features​

  • Supported field concatenation in data transformation and virtual columns.
  • Supported PolarDB-X 2.0 distributed database versions 5.4.21 and 8.4.
  • Added command whitelist filtering for Redis sources. New parameter: skipCommands.
  • Added support for setting the DataJob offset by timestamp for Dameng sources.
  • Supported data verification and data correction for SelectDB sources.
  • Added support for specifying the archive destination name when Oracle sources use LogMiner to read multiple archive log locations. New parameter: archiveDestName.
  • Added support for parsing source DDL with the Antlr-based parser for OceanBase for MySQL sources.
  • Added support for clearing target data before full migration or rerun for CloudBerry targets.
  • Added “Select All” support for batch operations on blacklist rules.
  • Added support for configuring target primary keys for TiDB and KingbaseES sources.
  • Added support for using force index as a query hint during TiDB full data queries.

Improvements​

  • Improved HANA CDC trigger data cleanup logic to avoid deleting unconsumed transaction data.
  • Improved the download and parsing process for RDS backup binlogs.
  • Improved performance when creating similar DataJobs in multi-database mode.
  • Improved StarRocks partition settings. When the target table already exists, the partition configuration is now disabled in the UI.
  • Optimized default parameters for Kafka sources and StarRocks targets to improve synchronization performance.
  • Improved MySQL, TiDB, PolarDB-X, PolarDB for MySQL, PostgreSQL, KingbaseES, and Vastbase sources by forcibly killing query processes after query failures.
  • During DataJob creation, the target data no longer fetches table PK/UK information.
  • Improved Dynamic mode for StarRocks, Doris, and SelectDB by fetching and parsing data in batches. New parameter: dynamicScanBatchSize.
  • Improved parsing of unknown type hints for schema and table names in PostgreSQL-compatible databases.
  • Improved frontend validation for Oracle sources in incremental + LogMiner mode when source table names exceed 30 characters.

Bug Fixes​

  • Fixed inaccurate source timestamp fields in HANA CDC output messages.
  • Fixed schema migration failures when the vector extension is enabled in a non-public schema in PostgreSQL.
  • Fixed an issue where JDBC cursor reads did not take effect when MySQL full migration or verification query filters contained semicolons, which could lead to OOM errors.
  • Fixed errors when KingbaseES targets executed UPDATE statements without a primary key and without data changes.
  • Fixed Jackson deserialization failures when RSocket or metadata payloads contained large fields over 20 MB.
  • Fixed an issue where, after changing target names during DataJob creation or when adding subscribed tables, the page status was not updated to “To be created” until a manual refresh.
  • Fixed false data verification mismatches for the PolarDB-X BIT type.
  • Fixed an issue where the Redis target parameter for skipping write errors did not take effect.
  • Fixed potential data loss caused by increasing LSN values after frequent restarts of incremental DataJobs for Dameng sources.
  • Fixed a dependency conflict with JDK 17.01 in Docker images that prevented the Console from starting properly.
  • Fixed an issue where DDL synchronization for Dameng sources incorrectly filtered DDL statements when the original SQL used lowercase table names.
  • Fixed an issue where incremental data for some tables could be filtered out when a single TiDB Region contained multiple tables, causing missing data on the target side.
  • Fixed an issue where subscription status could become inconsistent during TiDB Region split or Leader switch scenarios, preventing incremental data from being received normally.
  • Fixed an issue where old tableId or partitionId subscriptions were not cleaned up in time after TiDB source DDL or partition table changes, which could lead to abnormal states.
  • Fixed inaccurate time calculation for writing apply_commit.log commit logs after asynchronous writes to Kafka or Pulsar targets.
  • Fixed unsupported parser behavior when OceanBase for MySQL sources used SKIP_INDEX(SUM) type indexes.
  • Fixed a null pointer exception when Oracle sources in strongly typed incremental synchronization mode encountered null values from the source.
  • Fixed an issue where data correction DataJobs corrected columns that were not subscribed. New parameter: reviseAllSrcColumns.
  • Fixed inconsistencies between the actual page size limit and error messages in some paginated APIs.
  • Fixed an issue where the MySQL Antlr parser could not parse statements using backticks in options, such as row_format = dynamic``.
  • Fixed an issue where the Oracle Antlr parser could not parse statements such as alter table "DRAGONPASS"."TBLDRAGONDETAIL" shrink space CHECK;.
  • Fixed potential data loss in full synchronization for relational database sources when the source table had a primary key but the primary key column was not selected in wide-table column pruning scenarios.
  • Fixed an issue where pressing Enter did not confirm table creation.
  • Fixed an issue where, in expression mode, a DataJob could not be submitted after all existing expressions were deleted and new expressions were added during subscription modification.