deadlock - 使用很多`pool.Group`实例是否有开销（可能是递归的）？(Is there any overhead to using a lot of `pool.Group` instances (possibly in recursion)?)

Question

In a relatively large system using gevent I'm encountering a deadlock situation (the python process just freezes mid log, but no cpu is used).

(在使用gevent的相对较大的系统中，我遇到了死锁的情况（python进程仅冻结了中间日志，但未使用cpu）。)

I can't reproduce this reliably, but from the last few months it seems that the probability for this rises with the amount of nested pool.Group instantiations.

(我无法可靠地重现此错误，但是从最近几个月开始，这种可能性似乎随着嵌套pool.Group实例化的数量而pool.Group 。)

By nested I mean that I use the group to map some functions, and these functions I'm again doing a map using a new pool.Group .

(嵌套是指我使用组来映射一些函数，而这些函数又是使用新的pool.Group进行映射。)

Is there any limitation to creating pool.Group instances that I should be aware of?

(我应该注意的创建pool.Group实例是否有任何限制？)

Is there any difference from holding a global one and reusing it?

(与持有全球性和重复使用全球性有什么区别吗？)

(the nesting level is probably in the order of 10)

(（嵌套级别可能约为10）)

  ask by Uri translate from so

Answer 1

等待大神答复