v2.0.26

Data replication involves inherently complex and diverse workloads. Over the past several releases, we have continued to strengthen the foundation of Ape-DTS across three key areas: observability, through clearer metrics, error reporting, and runtime logs; controllability, through explicit resource configuration and the elimination of black-box behaviors that can unexpectedly amplify resource consumption; and usability, through streamlined configuration and CLI-based tooling. Alongside these foundational improvements, we have focused on making replication for four widely adopted open-source database engines—MySQL, PostgreSQL, MongoDB, and Redis—more stable and reliable across real-world workloads:

What's Changed

Added

General and Platform

• [471] - add configurable RPS and bandwidth limits for extractors and sinkers, and memory limits for pipeline buffers
• [460] - add inline CDC consistency checking
• [490] - add sampled consistency checking for snapshot and CDC workloads
• [500] - add dts-cli for configuration validation and database connectivity diagnostics
• [456] - support separate username and password fields when credentials contain URL-sensitive characters
• [544] - add busy-sinker metrics for worker utilization and scheduling visibility
• [525] - add profiling and Tokio Console image builds
• [530] - add runtime tracing for task scheduling and resource usage diagnostics

MySQL

• [450] - support parallel snapshot extraction using automatically selected or user-defined partition columns
• [483] - support TLS/SSL connections through configuration or URL parameters; certificate verification is not yet supported

PostgreSQL

• [450] - support parallel snapshot extraction using integer, floating-point, string, decimal, and other partition column types
• [475] - support selected user-defined functions and types, including enums, ranges, domains, and composite types, in structure migration
• [477] - support unchanged TOAST columns in CDC updates
• [483] - support TLS/SSL connections through configuration or URL parameters; certificate verification is not yet supported

Redis

• [508] - add Redis Cluster extraction support
• [511] - add database routing from source databases to target databases
• [520] - add Redis Cluster support to precheck and dts-cli

MongoDB

• [515] - support MongoDB 6.0+ Change Stream DDL replication
• [515] - support structure and CDC replication for sharded collections and shard keys
• [505] - support all valid MongoDB _id field types
• [520] - add MongoDB sharded-cluster support to precheck and dts-cli

Changed

General and Platform

• [553] - separate checker target connections, snapshot and CDC checks, and output settings into dedicated configuration sections
• [439] - improve checker difference logs, summary logs, and revision SQL output
• [442] - add consistency rechecking and improve review workflows
• [488] - replace snapshot multi-task fan-out with a single runtime task and internal scheduling
• [497] - add configurable snapshot chunk rebalancing based on row count or data size
• [538] - optimize sinker batch query construction and parameter binding
• [486] - expand integration and regression testing for supported databases and checker workflows

Redis

• [519] - automatically detect Redis Cluster mode instead of requiring explicit configuration

Breaking Changes

• [527] - remove the HTTP client/consumer and unmaintained Foxlake integrations to clarify product scope

Fixed

General and Platform

• [459] - fix failures when concurrent queues reach capacity
• [496] - fix a potential deadlock in task monitoring
• [521] - avoid duplicate pipeline capacity limiters
• [464] - fix DDL parsing when database or table names contain reserved keywords
• [509] - release metadata caches after table-level snapshot tasks and fix related routing issues
• [503] - fix RDB merge behavior when an update changes a unique key
• [530] - reduce high-frequency Tokio runtime polling and unnecessary wakeups
• [492] - fix validation failures caused by the default dummy-sinker configuration

MySQL

• [489] - fix large-packet transfers being restricted by the client's default packet limit
• [502] - fix compatibility with binary collations such as utf8_bin
• [469] - fix snapshot compatibility for certain time-related data types

PostgreSQL

• [473] - fix writes for custom types outside the search path or with duplicate names across schemas
• [472] - fix structure migration when comments contain single quotes or other special characters
• [474] - fix unique-index detection in metadata
• [476] - fix CDC pullback and concurrent merge behavior when primary or unique keys change
• [478] - fix CDC updates and deletes for tables without primary or unique keys
• [481] - fix parameter binding when an update condition contains NULL on a table without a primary key
• [550] - fix missing TRUNCATE events in CDC
• [467] - fix snapshot handling for "char" and other PostgreSQL data types

Redis

• [468] - fix PSYNC recovery positions not being restored correctly from logs
• [522] - fix handling of an empty is_cluster configuration value

MongoDB

• [547] - defer BSON parsing with RawDocumentBuf to handle invalid UTF-8 data and reduce extraction CPU usage