作业队列 - cron 可以工作吗？
我正在构建一个小型应用程序，要求人们通过电子邮件上传图像。它是用 PHP（无框架）、MySQL 和 S3 构建的。
到目前为止，在我的场景中：电子邮件存储在 POP3 帐户中。脚本每分钟运行一次，获取最旧的电子邮件，调整图像大小，将其上传到 S3，将其路径存储在数据库中，删除电子邮件。
在更大的范围内，如何管理？ cron 作业是处理这种情况的最佳方法吗？如果该过程花费超过一分钟怎么办：它会重叠并最终失败，对吗？或者如果不到一分钟怎么办？考虑到我每小时会收到超过 60 个更大规模的请求，我会得到不必要的空闲时间...
也许我应该使用 .forward 文件来处理电子邮件，但我同样不会控制流程。
我相信大多数场景都是有效的，我只是对最佳实践感到好奇。
谢谢！
I'm building a small application that requires people to upload images by email. It is built in PHP (no framework) with MySQL and S3.
So far, in my scenario: emails are stored on a POP3 account. A script runs every minute, fetches the oldest email, resizes the image, uploads it to S3, store its path in the DB, deletes the email.
In a larger scale, how would this be managed? Is the cron job the best way to handle this type of situation? What if the process takes more than a minute: it will overlap and eventually fail, right? Or what if it takes less than a minute? I'd get unwanted idle time considering I would have more than 60 requests an hour in a bigger scale...
Perhaps I should use a .forward file to process emails, but again I would not control the flow.
I believe most of these scenarios work, I'm just curious regarding best practices.
Thanks!
如果你对这篇内容有疑问，欢迎到本站社区发帖提问 参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（2）
稍微修改一下的方法可能是：
我曾经有一个备份程序，如果上次完成的备份早于一定时间，则每小时备份一次客户目录。这种方法非常有效，直到某些客户拥有太多 GB 的数据并且脚本需要一个多小时才能完成备份。
如果没有检查，下一小时脚本将再次运行同一客户，这也将花费一个多小时，依此类推，直到机器在非常高的运行级别下变得无响应。
我实施的修复是所描述的检查，如果另一个实例正在运行，只需退出并等待下一个周期。修复之后，我多年来从未遇到过任何问题。
A slightly modified approach could be:
I once had a backup procedure that was backing up customer directories every hour, if the last completed backup was older than a certain amount of time. This worked great until some customer had too many Gb of data and the script was taking more than a hour to do the backup.
Without the check, the next hour the script is going to run again the same customer, and that would take also more than a hour, and so on, until the machine becomes unresponsive with a very high runlevel.
The fix that I've implemented was the check described, if another instance was running, just exit and wait for the next cycle. After that fix I never had a problem for years.
尝试有一个长期运行的过程。它检查邮件并处理所有邮件。如果完成后找不到更多邮件，则会休眠一分钟。
如果您遇到稳定性问题，可以随时使用supervise之类的工具。
Try having a long-running process. It checks for mail, and processes all of it. If it can't find any more mail when it's done, then it goes to sleep for a minute.
If you have issues with stability, you can always use something like supervise.