关于多个worker同步upstream更新信息失败的问题

[huhaipeng]

看了代码，在更新upstream信息时是其中一个worker先主动更新，其他worker从共享内存处获取msg来同步更新upstream。若其中某些worker更新信息失败，造成不一致，应该怎么解决？我见代码里没有相应的解决方法。

[huhaipeng]

@yzprofile @chobits 大佬们可以解答一下小弟的疑惑嘛？

[jhseodev]

this module do not work normally when has worker processes more than two.
i want to know when will be updated.

[LucienCheng]

这个问题在多核多worker下出现了同步的问题，请问这个怎么解决呀~~~

[huhaipeng]

这个问题在多核多worker下出现了同步的问题，请问这个怎么解决呀~~~

你说的同步问题具体是什么问题？具体是怎么造成的？其实程序正常运行的情况下应该是不会出现同步的问题的吧，感觉和多核无关。

[huhaipeng]

this module do not work normally when has worker processes more than two.
i want to know when will be updated.

I test this module with multiple worker processes but nerver found any problems.So what is your specified problem ?

[jhseodev]

@huhaipeng
I'm using zone in upstream context as follow.

upstream my_ups {
  ...
  server ... ;
  zone ...;
  ...
}

location my_loc {
  ...
  set $dyups my_ups;
  proxy_pass http://$dyups;
  ...
}

When updating themy_ups upstream context, a new upstream context is created but zone is not initialized.
One worker process that received request is updated and the others are not updated.
This is what I have checked so far.

Please let me know if I am misunderstanding.
Thank for comment.

[dzhcool]

可以考虑在init_worker阶段给每个worker写个定时器，在定时器中处理upstream更新。