Skip to content

v1.1.0

Latest
Compare
Choose a tag to compare
@gaoyan1998 gaoyan1998 released this 30 Jul 14:03
2a13c50

Dinky-1.1.0 Release Note

incompatible changes

  • v1.1.0 supports the automatic schema upgrade framework (Flyway), using the table structure/data up to v1.0.2 as the default base version. If your version is not at v1.0.2+, you must first upgrade to the table structure of v1.0.2 according to the official upgrade tutorial. If your version is v1.0.2+, you can directly upgrade, and the program will automatically execute without affecting historical data. If you are deploying from scratch, please ignore this matter.

  • Due to the contribution of flink-cdc to the Apache Foundation, the package name of the new version will change, and it is not possible to make compatibility changes. In versions dinky-v1.1.0 and above, dinky will use new package name dependencies, which requires your flink-cdc dependencies to be upgraded to flink-cdc v3.1+, otherwise it will not work.

  • Remove the distinction of Scala version when packaging, only develop with Scala-2.12, and no longer support Scala-1.11.x.

New Features

  • Added Flyway schema upgrade framework.
  • Task directory supports flexible sorting.
  • Implemented task-level permission control and supports different permission control strategies.
  • Optimized the automatic addition of administrator user association when adding tenants.
  • Added the function to directly kill the process in the case of task submission deadlock.
  • Support k8s deployment of dinky.
  • Implement data preview.
  • New support for UDF injection configuration in data development.
  • Added whole library synchronization function (cdcsource) sink end table name mapping, regular matching modification mapping.
  • Added Dashboard page.
  • Added Paimon data source type.
  • Added SQL-Cli.

Fixes

  • Modified the issue with k8s's account.name value and added the problem of Conf initialization when deleting a cluster.
  • Fixed the issue of flink-cdc losing SQL in application mode.
  • Fixed the issue where the task creation time was not reset when copying tasks.
  • Fixed the task list positioning problem.
  • Solved the problem of user-defined classes in user Jars not being compiled when submitting Jar tasks.
  • Fixed the incorrect alarm information in the enterprise WeChat-app mode.
  • Fixed the problem of flink-1.19 not being able to submit tasks.
  • Fixed the startup script not supporting jdk11.
  • Fixed the problem of cluster instances not being deleted.
  • Fixed the problem of UDF not finding the class in Flink SQL tasks.
  • Fixed the problem of the data development page not updating the state when the size changes.
  • Fixed the problem of not being able to get the latest high availability address defined in custom configuration.
  • Fixed the problem of not recognizing the manual configuration of rest.address and rest.port.

Optimizations

  • Optimized the prompt words in resource configuration.
  • Optimized the DDL generation logic of the MySQL data source type.
  • Optimized some front-end dependencies and front-end prompt information.
  • Optimized the copy path function of the resource center, supporting multiple application scenarios within dinky.
  • Optimized the monitoring function, using the monitoring function switch in dinky's configuration center to control all monitoring within dinky.
  • Optimized some front-end judgment logic.

Restructuring

  • Moved the alarm rules to the alarm route under the registration center.
  • Removed paimon as the storage monitoring medium, changed to sqllite, and do not strongly depend on the hadoop-uber package (except in the Hadoop environment), and support periodic cleaning.
  • Restructured the monitoring page, removing some built-in service monitoring.

Documentation

  • Added documentation for deploying dinky on k8s.
  • Optimized the Docker deployment documentation.
  • Added documentation related to whole library synchronization function (cdcsource) sink end table name mapping.