BladePipe 1.10.0: Support Google Ads and more, improve SAP HANA source CDC performance.
Skip to main content

BladePipe 1.10.0

Release Date: August 31, 2026

Version: 1.10.0

Highlights​

  • SAP HANA sources now support non-blocking handling of uncommitted transactions during table-level CDC. Changes are automatically replayed after the transaction is committed or rolled back. (New parameter: uncommittedTxMode)
  • PostgreSQL 17+ sources now support failover-enabled replication slots, allowing CDC to continue after a primary/standby switchover without rebuilding the slot.
  • Oracle CDC adds a new ADD_FILE_ARCHIVE_WINDOW mode that dynamically includes additional archive logs around the current mining range to provide sufficient transaction context. Queries are also narrowed to the actual SCN range for better LogMiner performance.

New Pipelines​

  • Google Ads > MySQL: Incremental sync via scheduled scans with upsert support.
  • PostgreSQL > PolarDB for MySQL: Schema migration, full migration, CDC, data verification, data correction, and DDL sync, including add/drop/modify/rename columns and truncate operations.
  • TDSQL MySQL / TDSQL-C MySQL > StarRocks / Doris: Full migration, CDC, data verification, data correction, schema migration, and DDL sync, including add/drop/modify/rename columns and truncate operations.
  • TDSQL MySQL / TDSQL-C MySQL > Kafka: Full migration and CDC.
  • Dameng / KingbaseES / GaussDB > Elasticsearch: Full migration, CDC, data verification, and schema migration.

New Features​

  • Doris > Oracle scheduled incremental sync can now use DATETIME or DATETIMEV2 columns as periodic filter fields.

  • Doris targets now support column default values during schema migration and DDL sync for supported numeric and string types.

  • OceanBase for MySQL > OceanBase for MySQL schema migration now preserves table partition definitions.

  • PostgreSQL sources now support the non-intrusive CHECK_POS heartbeat mode.

  • SQL Server target verification automatically splits large queries to avoid exceeding parameter limits. (New parameter: maxInSizePerQuery)

  • MySQL > SQL Server DDL sync now skips index and comment operations that have already been applied during pre-checks, preventing duplicate DDL execution errors on the target.

  • Log downloads now use a dedicated communication channel, isolated from critical operations such as DataJob scheduling for improved stability.

  • Live log viewing (tail) now uses a dedicated communication channel, improving stability without affecting DataJob scheduling.

  • Console monitoring metrics collected from Sidecars now use a dedicated communication channel, isolated from critical DataJob operations.

  • MFA now supports custom user and service identifiers.

Improvements​

  • Improved SAP HANA trigger-check performance.

  • Creating a similar DataJob from an SAP HANA source now preserves advanced configuration settings.

  • Improved Kafka source offset lookup compatibility. Brokers earlier than Kafka 0.11 can now retrieve partition end offsets and committed consumer-group offsets.

  • Improved Kafka source offset reset behavior:

    • Supports retries for active consumer groups.
    • Automatically falls back to the latest offset when a timestamp is out of range.
    • Fixes an incorrect timeout unit.

  • Oracle CDC now exposes the number of remaining archive log files to be processed.

  • Doris full migration now retrieves row counts from metadata instead of scanning the entire table.

  • StarRocks full migration now retrieves row counts from metadata instead of scanning the entire table.

  • Improved dstWholeReplace writes for MySQL, ADB for MySQL, PolarDB for MySQL, and TDSQL-C MySQL targets by splitting incremental events based on column structure.

  • Improved KingbaseES compatibility when system search_path settings prevent direct access to system tables.

  • KingbaseES now explicitly converts index ordinal types for compatibility with strict function-argument matching.

  • Dameng CDC now filters DML events with empty table names from LogMiner output.

  • Elasticsearch target verification now ignores property order in OBJECT fields.

  • GaussDB adds replication-stream heartbeats to prevent WAL sender disconnections during processing stalls. (New parameter: replicationHeartbeatSec)

  • Improved log downloads with streaming support so large files no longer block the page.

  • DataSource endpoint, username, password, and related settings are now managed from a single edit entry point.

  • Resetting AK/SK now requires secondary confirmation to reduce accidental operations.

  • Full migration filter pushdown now applies only to tables with configured filters. Tables without filters continue to use paginated scanning.

  • Async DataJob operation records now show the exact sub-account username.

  • Improved DataJob parameter templates.

  • BladePipe now warns users when creating a DataJob for a table without a primary key if no target primary key has been configured.

Bug Fixes​

  • Upgraded Fastjson in DataJob runtime components to version 1.2.84 to address a security issue.

  • Fixed an issue where SQL Server CDC could miss changed columns when the CDC update mask did not correctly flag them.

  • Fixed an issue where removed Kafka topics could remain in stored offsets after updating subscriptions for multi-offset DataJobs.

  • Fixed an issue where Paimon, Delta Lake, and Iceberg full or incremental writes could not skip failed records according to DataJob settings. Hudi full-load asynchronous flush errors can now also be skipped.

  • Fixed PostgreSQL > MySQL interval type mapping. It is now mapped to varchar.

  • Fixed nullable constraints on target primary-key-related columns when writing to OceanBase for Oracle.

  • Fixed JSONB escape parsing for MySQL.

  • Fixed DROP TABLE DDL sync compatibility with SQL Server versions earlier than 2016.

  • Fixed PostgreSQL > MySQL-family pipelines where:

    • Boolean values could be lost.
    • Floating-point arrays containing NaN or Infinity could fail to write.

  • Fixed an issue where PostgreSQL target DDL sync could unintentionally add a NOT NULL constraint when changing a column type.

  • Fixed replication slots not being removed after KingbaseES/Vastbase sub-DataJobs were merged.

  • Fixed TDSQL-C MySQL CDC failing to locate binlog files after a primary/standby switchover.

  • Fixed incorrect fractional-second precision detection for Dameng TIMESTAMP WITH LOCAL TIME ZONE caused by undecoded DATA_SCALE metadata.

  • Fixed GaussDB CDC losing the sign of MONEY values.

  • Fixed MySQL, PolarDB for MySQL, and TDSQL-C MySQL batch writes where TINYTEXT, TEXT, MEDIUMTEXT, or LONGTEXT values could be written as empty strings when the character-stream length was not specified.

  • Fixed a Dameng full migration SQL syntax error that could occur when filter pushdown was enabled but no filter condition was configured.

  • Fixed PostgreSQL > MySQL full migration writing empty arrays as "]".

  • Fixed loss of table comments during MySQL → SQL Server DDL sync.

  • Fixed an issue where removing the final alert method from DataJob alert settings did not take effect.

  • Fixed stale values in the database-mapping dropdown when creating a similar DataJob.

  • Fixed an error when adding multiple rows at once and selecting databases one by one in multi-database DataJob creation.

  • Fixed quoted field names in filter conditions not being recognized correctly.

  • Fixed an incorrect “To be created” status when updating a subscription and first selecting a target column that did not exist while the target table itself already existed.

  • Fixed an issue where rapidly closing multiple error dialogs could cause some error messages to be lost.