delayed_job 的 基本用法
delayed_job 是用来做某些可以延后的，对时间要求不高的任务，例如： 发送1W个邮件，处理10W个数据，不能在规定的时间内（例如1秒）立即执行完的任务。
delayed_job 有两个版本，一个是最初的：
https://github.com/tobi/delayed_job
另一个是改进的：
https://github.com/collectiveidea/delayed_job
我用的是第二个。它们用法几乎一样，除了在 优先级方面(priority )，后者是数字越大，优先级越低。
文档方面很奇怪，建议两者都要看，互相综合，才能看完整。
下面把我的一点心得贴出来：
1. worker 至关重要。默认是启动一个，看起来是单线程。 启动时配置参数：
$ script/delayed_job -n 4 start (启动4个worker)
2. 每个job一旦被执行，那么它被lock住，(lock=true)，等到该JOB彻底被执行完毕，才会把它从delayed_job 表中删除掉。
3. 可以在 config/initializers 中进行配置：
# delayed_job_config.rbDelayed::Worker.destroy_failed_jobs = trueDelayed::Worker.sleep_delay = 30Delayed::Worker.max_attempts = 5 Delayed::Worker.max_run_time = 10.minutes
def invoke_job hook :before payload_object.perform hook :success rescue Exception => e hook :error, e raise e ensure hook :after end
class ParanoidNewsletterJob < NewsletterJob def perform emails.each { |e| NewsletterMailer.deliver_text_to_email(text, e) } end def before(job) record_stat 'newsletter_job/start' endendclass EatFoodJob < Struct.new(:job_name) def perform Rails.logger.info "===(#{job_name}) running" sleep_time = 1 + rand(5) sleep sleep_time Rails.logger.info "===(#{job_name}) after #{sleep_time} sleep, done " end endclass WaitJob < Struct.new(:job_name) def perform Rails.logger.info "===(#{job_name}, #{self.class}) running" loop do break unless DelayedJob.where(:priority => 10).all.size > 0 Rails.logger.info " === !!!!!! jobs: priority => 10 not done, now wait for 5 seconds" sleep 1 end Rails.logger.info "===(#{job_name}) done " end endclass DrinkWaterJob < Struct.new(:job_name) def perform Rails.logger.info "===(#{job_name}, #{self.class}) running" sleep_time = 5 + rand(5) sleep sleep_time Rails.logger.info "===(#{job_name}) after #{sleep_time} sleep, done " end end(1..10).each do |i| Delayed::Job.enqueue DrinkWaterJob.new("drink#{i}"), :priority => 10 Delayed::Job.enqueue EatFoodJob.new("eat#{i}"), :priority => 20end(1..4).each do |i| Delayed::Job.enqueue WaitJob.new("wait#{i}"), :priority => 15end