BladePipe 1.9.0: New data pipelines, faster Oracle writes, and improved stability.
跳到主要内容

BladePipe 1.9.0

Release Date: July 31, 2026

Version: 1.9.0

Highlights​

  • Support staging-table merge writes to Oracle targets during incremental synchronization, significantly improving target-side write performance. New parameters: useStagingMerge and stagingSchema.
  • Support staging-table merge writes to OceanBase for Oracle targets during full migration and incremental synchronization, significantly improving target-side write performance. New parameters: useStagingMerge and stagingSchema.

New Pipelines​

  • PostgreSQL -> Oracle: schema migration, full migration, incremental synchronization, data verification, data correction, and DDL synchronization (ADD COLUMN, DROP COLUMN, MODIFY, RENAME, and TRUNCATE).
  • Iceberg -> Redshift / MySQL: schema migration, full migration, and incremental synchronization through scheduled scans and upserts.
  • MySQL / PostgreSQL / Aurora PostgreSQL / SQL Server -> Redshift: schema migration, full migration, and incremental synchronization.
  • PostgreSQL / Aurora PostgreSQL / SQL Server -> Iceberg: schema migration, full migration, and incremental synchronization.

New Features​

  • Support Oracle 21c, 23c, 23ai, and 26ai.
  • Support periodic incremental synchronization from SshFile / OssFile / S3File / GoogleDrive / Yuque. Files are migrated automatically when changes are detected, and documents that no longer exist at the source are skipped by default.
  • Added support for automatically setting source-table primary keys as Identifier Fields on Iceberg targets during schema migration.
  • Support scheduled-scan incremental synchronization from Dameng to StarRocks.
  • Added case-insensitive automatic mapping of target tables on the table-mapping page.
  • Added support for configuring the initial consumption strategy for Kafka sources. Available options: earliest and latest. New parameter: kafkaAutoOffsetReset.
  • Added support for configuring database heartbeat mode for PolarDB MySQL and PolarDB-X sources. New parameter: dbHeartbeatMode.
  • Added support for filtering alert logs by notification method.
  • Added support for parsing TRUNCATE events from PostgreSQL / KingbaseES / Vastbase sources.
  • Added support for synchronizing tables without a primary key from StarRocks / Doris / SelectDB, but you have to specify the target primary key mannually.

Improvements​

  • Improved the replication-stream heartbeat mechanism for PostgreSQL / KingbaseES / Vastbase sources to prevent disconnections due to WAL Sender timeout when DataJob processing is blocked. New parameter: replicationHeartbeatSec.
  • Fixed write-performance degradation caused by ineffective batched INSERT execution on OceanBase for Oracle targets.
  • Improved Oracle target write performance by resolving unstable JDBC binding types that caused batched INSERT / MERGE execution to become ineffective.
  • Improved the performance of locating Binlog positions by timestamp for MySQL, OceanBase for MySQL, PolarDB MySQL, PolarDB-X, TDSQL-C MySQL, and TDSQL MySQL sources. New parameter: fastTimestampSeekEnabled, disabled by default.
  • Improved metadata retrieval performance for TiDB, ADB MySQL, GreptimeDB, PolarDB, TDSQL MySQL, TDengine, StarRocks, DeltaLake, Iceberg, and Paimon.
  • Improved subscription updates so that child DataJobs inherit the keyConflictStrategy parameter from the parent DataJob.
  • Improved full-scan support for MySQL 5.7 views.
  • Improved phone alert logs by adding DataJob information.
  • Improved the Add DataSource workflow so that sensitive fields are displayed in encrypted form.
  • Added a prompt in the Create DataJob and Modify Subscription workflows when not all columns are selected.
  • Improved the Create Similar DataJob workflow by preventing users from proceeding until database information has finished loading.
  • Improved duplicate-click prevention for buttons across the system.
  • Improved the “Search and select manually” mode so that previously selected items remain selected after another filter is applied.
  • Improved the “Search and select manually” mode to support filtering both source and target tables at the same time.

Bug Fixes​

  • Fixed an issue where the DataJob position was not updated when synchronizing incrementally from Kafka to StarRocks in multi-position mode.
  • Fixed duplicate accumulation of synchronized row counts when updating positions across multiple DynamoDB shards.
  • Fixed table-name search filtering for Kafka, RabbitMQ, and other messaging pipelines in “List all and select manually” mode.
  • Fixed an issue in the Create DataJob workflow where switching from expression mode to another mode after returning to the previous step caused the data-cleaning step to be skipped.
  • Fixed an issue where switching to expression mode and then switching back during DataJob creation could result in a configuration error after the DataJob was created successfully.
  • Fixed an issue where modifying the GTID did not take effect when resetting the position of a MySQL source.
  • Fixed failures when parsing Date fields in epoch_millis or epoch_second format from Elasticsearch sources.
  • Fixed an issue where PostgreSQL enum types were not recognized, preventing data synchronization.
  • Fixed parsing errors when synchronizing BC date values from PostgreSQL sources.
  • Fixed full synchronization failures when PostgreSQL DECIMAL fields contained Infinity, -Infinity, or NaN.
  • Fixed an issue where an IP address could not be obtained when all host network interfaces were in bridge mode. Standard network interfaces are now preferred, with bridge-interface IPv4 addresses used only when no valid standard-interface IP is available.
  • Fixed an issue where executing DDL to modify a primary-key type or comment on MySQL / OceanBase for MySQL sources could overwrite and remove primary-key information from real-time metadata, causing subsequent data inconsistencies.
  • Fixed incomplete file listing when an S3File bucket contained more than 1,000 files.
  • Fixed an issue where an incorrect verification code did not produce the proper prompt when resetting MFA.
  • Fixed abnormal decimal values when Oracle full migration used getString to read NUMBER values with nonstandard encoding.
  • Fixed inconsistent primary-key query order when performing data validation, rechecks, or data correction on tables with composite primary keys.
  • Fixed validation SQL failures when a primary-key column name was a database keyword.
  • Fixed an issue where heartbeat messages were filtered out in PARSED_ENTRY mode for MySQL, OceanBase for MySQL, PolarDB MySQL, and PolarDB-X sources, causing heartbeats to become ineffective.
  • Fixed an issue where CHECK_POS heartbeats for MySQL, OceanBase for MySQL, PolarDB MySQL, and PolarDB-X compared only the Binlog offset, causing identical offsets in different Binlog files to be incorrectly treated as zero lag.
  • Fixed failures when constructing position heartbeats for MySQL, OceanBase for MySQL, PolarDB MySQL, and PolarDB-X sources in PROTOBUF non-GTID mode when the GTID was empty.
  • Fixed a potential error in locating the transaction start position when MySQL, OceanBase for MySQL, PolarDB MySQL, PolarDB-X, TDSQL-C MySQL, and TDSQL MySQL sources resumed incremental synchronization from a saved position.
  • Fixed an issue where active Binlog download requests and download threads did not exit promptly when stopping a DataJob with a MySQL RDS source.
  • Fixed MCP tool invocation failures when BladePipe was deployed behind a load-balancing proxy.
  • Fixed missing spaces after separators in TiDB JSON data during incremental synchronization.
  • Fixed occasional ConcurrentModificationException errors caused by a thread-unsafe TableUnit cache during concurrent Kafka parsing on JDK 17.
  • Fixed Arrow MemoryUtil initialization failures during StarRocks full reads on JDK 17.
  • Fixed DataJob startup failures for PolarDB-X and OceanBase for MySQL caused by configuration compatibility issues.
  • Fixed schema migration syntax errors caused by retaining DEFAULT values when mapping TEXT / BLOB columns to MySQL-compatible targets.
  • Fixed an issue where unsupported statements in Oracle-source DDL interrupted incremental synchronization. Such statements can now be identified and skipped correctly.
  • Fixed an issue where the Elasticsearch index metadata whitelist did not take effect during DataJob creation.
  • Fixed Sidecar startup failures during automatic upgrades or installation when the clougence user did not exist on the target machine.
  • Fixed inaccurate process PID detection during automatic Sidecar upgrades, which could prevent Sidecars or DataJobs from stopping correctly.
  • Fixed Sidecar startup failures when automatic upgrades or deployments were executed with the root account and the clougence user’s environment variables were not inherited.