Ruby 3 + Async 2.3.0中无法捕获MetaInspector::TimeoutError的排查求助

你的代码里rescue块没生效，核心原因是Async的IO调度器和MetaInspector的超时机制不兼容：

• MetaInspector内部依赖的HTTP客户端（比如Net::HTTP）使用Ruby原生Timeout实现超时，但Async的调度器会接管所有IO操作，超时错误会从Async的事件循环（async/scheduler.rb的run_once方法）抛出，这个异常不在你写的begin/rescue代码块的上下文里。
• 嵌套Async任务的异常如果没被任务内部完全处理，会直接冒泡到顶层Async任务，导致程序崩溃，而不会被你写的rescue捕获。

方案1：使用Async原生的超时机制（推荐） #

放弃MetaInspector的超时配置，改用Async提供的Async::Timeout来控制任务超时，这样异常能被正常捕获：

require 'async'
require 'async/timeout'
require 'metainspector'

urls = ["https://httpbin.org/delay/15", "https://www.google.com"]

Async do 
  urls.each do |url|
    Async do
      begin
        # 设置总超时15秒（包含连接+读取）
        Async::Timeout.timeout(15) do
          page = MetaInspector.new(url, allow_non_html_content: true)
          # 可以在这里添加对page的处理逻辑
        end
        puts "made it to the end of the nested Async block with #{url}"
      rescue Async::Timeout::Error
        puts "任务超时: #{url}"
      rescue => e
        puts "rescued everything else, type: #{e.class}, message: #{e.message}"
      end
    end
  end
end

方案2：捕获Async任务的全局异常 #

给每个Async任务绑定on_error回调，确保任务抛出的所有异常都能被捕获：

require 'async'
require 'metainspector'

urls = ["https://httpbin.org/delay/15", "https://www.google.com"]

Async do 
  urls.each do |url|
    task = Async do
      begin
        page = MetaInspector.new(url, allow_non_html_content: true)
        puts "made it to the end of the nested Async block with #{url}"
      rescue MetaInspector::TimeoutError
        puts "rescued MetaInspector's TimeoutError"
      rescue => e
        puts "rescued everything else, type: #{e.class}, message: #{e.message}"
      end
    end

    # 捕获任务未处理的异常
    task.on_error do |error|
      puts "任务未捕获异常: #{error.class} - #{error.message}"
    end
  end
end

方案3：让MetaInspector使用兼容Async的HTTP客户端 #

MetaInspector支持自定义HTTP客户端，你可以配置它使用async-http作为适配器，这样超时逻辑会和Async调度器兼容，异常能被正常rescue捕获。需要额外安装相关依赖，配置示例：

# 先安装依赖：gem install async-http faraday-async-http
require 'async'
require 'metainspector'
require 'faraday'
require 'faraday/async_http'

# 配置MetaInspector使用Async HTTP客户端
MetaInspector.configure do |config|
  config.faraday_adapter = :async_http
end

urls = ["https://httpbin.org/delay/15", "https://www.google.com"]

Async do 
  urls.each do |url|
    Async do
      begin
        page = MetaInspector.new(url, connection_timeout:10, read_timeout:5, allow_non_html_content: true)
        puts "made it to the end of the nested Async block with #{url}"
      rescue MetaInspector::TimeoutError
        puts "rescued MetaInspector's TimeoutError"
      rescue => e
        puts "rescued everything else, type: #{e.class}, message: #{e.message}"
      end
    end
  end
end

内容的提问来源于stack exchange，提问作者TepidSwitch