Rails 4应用POST请求后偶发冻结：ActiveRecord及缓存排查求助

Let's break down what's causing your app to freeze for 5+ minutes and how to fix it. Looking at your logs, code, and the intermittent, hard-to-reproduce symptoms, there are a few critical areas to address:

1. External API Calls Are Likely Hanging #

Your create action runs report_play_to_shoutcast and report_play_to_tunein synchronously during the request cycle. If these third-party API requests don't have explicit timeouts configured, they can hang for minutes when the external service is slow or unresponsive. This ties up a Rails process/thread, blocking all other user requests until the call finally times out—exactly the site-wide freeze you're experiencing.

Fixes: #

• Add strict timeouts to all external HTTP calls. For example, if using Net::HTTP, cap the connection and read timeouts to 5 seconds max:

  def report_play_to_shoutcast(play)
    uri = URI("your-shoutcast-api-endpoint")
    http = Net::HTTP.new(uri.host, uri.port)
    http.open_timeout = 5
    http.read_timeout = 5
    # Rest of your API request logic here
  end
  

• Move these calls to a background job (like Sidekiq or Resque). This way, the user's request finishes immediately, and the API calls run in the background where they can't block the main application.

2. Cache Bulk Deletion Is Blocking Your Cache Service #

Your clear_plays_cache method uses Rails.cache.delete_matched('plays__'). On cache stores like Redis, this operation scans every matching key to delete them—and if you have a large number of cached entries, this can take a long time and block the entire cache service. Any request that relies on cache (like last_play or plays_count) will hang until the deletion completes.

Fix: Use Cache Versioning Instead #

Replace slow bulk deletion with a versioned cache key strategy. This is atomic, fast, and avoids blocking:

# In ApplicationHelper
def plays_cache_version
  Rails.cache.fetch("plays_cache_version", expires_in: 1.year) { SecureRandom.uuid }
end

# Update cache keys to include the version
def last_play_cache_key
  "plays__last_play_v#{plays_cache_version}"
end

def set_last_play(play)
  Rails.cache.write(last_play_cache_key, play, expires_in: 5.minutes, race_condition_ttl: 10.seconds)
end

def last_play
  @last_play ||= Rails.cache.fetch(last_play_cache_key, expires_in: 5.minutes, race_condition_ttl: 10.seconds) do
    Play.order("created_at").includes(:show).last || Play.new(album: "", artist: "", track: "", show: nil)
  end
end

# Now clearing the cache just updates the version (no bulk deletion needed)
def clear_plays_cache
  Rails.cache.write("plays_cache_version", SecureRandom.uuid)
end

This way, all old cache keys are automatically ignored when the version changes—no slow scans or blocking operations required.

3. Optimize Database Queries in set_plays #

Your set_plays method uses Play.all.paginate(...)—while will_paginate adds LIMIT/OFFSET, the ORDER BY created_at DESC can be slow if you don't have an index on the created_at column. Additionally, Play.all is redundant in Rails 4 (you can directly use Play to start the query chain).

Fixes: #

• Add an index on created_at to speed up the sort operation:

  # Generate a migration with: rails generate migration add_index_to_plays_created_at
  add_index :plays, :created_at
  

• Clean up the query syntax:

  def set_plays
    @plays = Play.order('created_at DESC').preload(:show,:playbox)
                 .paginate(page: params[:page], per_page: 50, total_entries: plays_count())
  end
  

• Optimize plays_count: Instead of relying on cache expiration to trigger a slow Play.count, update the count in a background job (e.g., every minute) so the cache stays fresh without blocking requests.

4. Debug to Confirm the Culprit #

To pinpoint exactly where the time is being spent, add detailed timing logs around suspicious code blocks:

# In your create action
start_time = Time.now
report_play_to_shoutcast(@play) if Rails.env.production?
Rails.logger.info "Shoutcast report took #{Time.now - start_time} seconds"

start_time = Time.now
clear_plays_cache
Rails.logger.info "Cache clear took #{Time.now - start_time} seconds"

This will show you which step is responsible for the 5-minute freeze.

内容的提问来源于stack exchange，提问作者jasonsogan