Release v3.1.0

v3.1.0

发布总目录 · 版本与升级 · GitHub Release

• 类型：框架正式版
• 发布日期：2023-07-27T12:32:24Z
• Release 标题：v3.1.0

以下为 GitHub 发布说明原文。它与主仓库 CHANGELOG 分别保存；如有差异，请核对目标 tag 的源码。

Breaking Changes

For users upgrading from 3.0.x or below, please notice that the default service discovery model in 3.1.0 has changed to registry-type: instance, which stands for the application-level discovery model in Dubbo3.

To be able to discover instances of lower version, please set registry-type of 3.1.0 instance to interface in via dubbogo.yml or API:

dubbo:
  registries:
     zk: 
       protocol: zookeeper
       address: 127.0.0.1:2181
       registry-type: interface

Features

• Implement RemoveConfig API for nacos and zookeeper
• Expose TLSConfig for config api
• Add subscribe any value
• Implement condition routing basic functions and complete related tests
• Implement dynamic routing
• Tag route static config
• Upgrade nacos 2.x
• Print logs of registered providers and consumers
• Add request processing total and request succeed total metrics
• Add some metrics about RT
• Add metrics base api interface
• Sliding window and rt quantile metrics

Bugfixes

• Fix: Call Java provider occur error use group+version on triple protocol
• Fix: Register instance protocol info to polaris when use Tripe
• Fix: Custom triple server message size
• Fix: Start a new routine when subscribing
• Fix: Upgrade hessian2 to fix java wrapper object decoding bug
• Fix: Polaris service discovery cause nil panic
• Fix: The metadata invoker is destroyed too early
• Fix: Format imports in metrics module
• Fix: Solve config bool field zero value bug by using pointer
• Fix: Disable metrics filter by default instead

Enhancements

• Add serviceMappingListener for application service discovery
• The argument to ioutil.ReadFile in the config.WithPath method is changed to conf.path
• Upgrade hessian2 to support java wrapper types
• Tablewrite cli show
• Refactor: remove providers and add prefix when use polaris
• Log configuration supports other frameworks and can configure log file generation directory and file size
• Ignore node not exist error message and continue retry
• Refactor: Split metrics module into multiple files
• Ignore zk node already exist when store provider/consumer metadata
• Try update value if node already exists
• Refactor: Use promauto and promhttp
• Chore: Unnecessary use of fmt.Sprintf
• Chore: Pkg imported more than once