捕获特定错误并重试

Question

我有一个ruby on rails应用程序，它每小时执行一次rake tweet，从数据库中发出一条推文。我注意到有几次推特还没有发送，在查看日志后我发现：

2013-02-11T19:00:39+00:00 app[scheduler.8796]: Connecting to database specified by  DATABASE_URL
2013-02-11T19:00:39+00:00 app[scheduler.8796]: (__-){ Twitter is over capacity.
2013-02-11T19:00:39+00:00 app[scheduler.8796]: /app/vendor/bundle/ruby/1.9.1/gems/twitter-4.5.0/lib/twitter/response/raise_error.rb:21:in `on_complete'
2013-02-11T19:00:39+00:00 app[scheduler.8796]: /app/vendor/bundle/ruby/1.9.1/gems/faraday-0.8.5/lib/faraday/response.rb:9:in `block in call'
2013-02-11T19:00:39+00:00 app[scheduler.8796]: rake aborted!
2013-02-11T19:00:39+00:00 app[scheduler.8796]: /app/vendor/bundle/ruby/1.9.1/gems/faraday-0.8.5/lib/faraday/response.rb:8:in `call'
2013-02-11T19:00:39+00:00 app[scheduler.8796]: /app/vendor/bundle/ruby/1.9.1/gems/faraday-0.8.5/lib/faraday/response.rb:8:in `call'
2013-02-11T19:00:39+00:00 app[scheduler.8796]: /app/vendor/bundle/ruby/1.9.1/gems/faraday-0.8.5/lib/faraday/response.rb:63:in `on_complete'
2013-02-11T19:00:39+00:00 app[scheduler.8796]: /app/vendor/bundle/ruby/1.9.1/gems/faraday-0.8.5/lib/faraday/request/multipart.rb:13:in `call'
2013-02-11T19:00:39+00:00 app[scheduler.8796]: /app/vendor/bundle/ruby/1.9.1/gems/faraday-0.8.5/lib/faraday/response.rb:8:in `call'
2013-02-11T19:00:39+00:00 app[scheduler.8796]: /app/vendor/bundle/ruby/1.9.1/gems/faraday-0.8.5/lib/faraday/connection.rb:110:in `post'
2013-02-11T19:00:39+00:00 app[scheduler.8796]: /app/vendor/bundle/ruby/1.9.1/gems/twitter-4.5.0/lib/twitter/request/multipart_with_file.rb:14:in `call'
2013-02-11T19:00:39+00:00 app[scheduler.8796]: /app/vendor/bundle/ruby/1.9.1/gems/faraday-0.8.5/lib/faraday/connection.rb:245:in `run_request'
2013-02-11T19:00:39+00:00 app[scheduler.8796]: /app/vendor/bundle/ruby/1.9.1/gems/faraday-0.8.5/lib/faraday/request/url_encoded.rb:14:in `call'
2013-02-11T19:00:39+00:00 app[scheduler.8796]: /app/lib/tasks/scheduler.rake:11:in `block in <top (required)>'
2013-02-11T19:00:39+00:00 app[scheduler.8796]: /app/vendor/bundle/ruby/1.9.1/gems/twitter-4.5.0/lib/twitter/api/tweets.rb:129:in `update'
2013-02-11T19:00:39+00:00 app[scheduler.8796]: /app/vendor/bundle/ruby/1.9.1/gems/twitter-4.5.0/lib/twitter.rb:52:in `method_missing'
2013-02-11T19:00:39+00:00 app[scheduler.8796]: /app/vendor/bundle/ruby/1.9.1/gems/twitter-4.5.0/lib/twitter/api/utils.rb:82:in `object_from_response'
2013-02-11T19:00:39+00:00 app[scheduler.8796]: /app/vendor/bundle/ruby/1.9.1/gems/twitter-4.5.0/lib/twitter/client.rb:81:in `request'
2013-02-11T19:00:39+00:00 app[scheduler.8796]: /app/vendor/bundle/ruby/1.9.1/gems/twitter-4.5.0/lib/twitter/client.rb:70:in `post'
2013-02-11T19:00:39+00:00 app[scheduler.8796]: (See full trace by running task with --trace)
2013-02-11T19:00:39+00:00 app[scheduler.8796]: Tasks: TOP => tweet
2013-02-11T19:00:41+00:00 heroku[scheduler.8796]: Process exited with status 1

我如何才能捕捉到这个特定的错误并重试推文，直到它被发送？

在我的rake文件中：

task :tweet => :environment do
 if ApprovedTweet.all.size > 0
    t = ApprovedTweet.first
    Twitter.update(t.text)
    t.destroy
 end

Answer 1

从Twittergem的文档中可以看出，Twitter::Error::TooManyRequests似乎是您试图捕捉的错误。

如果我是你，我不会在推特发送之前重试发送推特（尤其是在rake任务中，你很容易会陷入僵尸进程），而是会允许在失败前多次重试：

begin     
  Twitter.update(t.text)
rescue Twitter::Error::TooManyRequests => e
  retries ||= 0
  retries +=  1
  if retries < ARBITRARY_MAX_RETRY_LIMIT
    # optionally sleep or delay for a while
    # ... or just log failure 
    retry
  else
    raise e # or another custom error of yours
  end
end