init_master patch?

[Maxim Dounin]

Hello!

On Fri, May 20, 2016 at 11:24:33AM -0700, Stephan Zuercher wrote:

> Hi,
> 
> I see in the past there's been questions about ngx_module_t's init_master
> field never being called.
> 
> I'm working on a plug-in that needs to execute code in the context of the
> master process (after it's forked to become a daemon) and init_master seems
> like the logical way to do it.
> 
> If I contribute a change to call init_master, is it likely to be accepted?
> Or is there some fundamental reason not to do this?

It's believed that init_master is a wrong thing to use in most 
cases.  In particular, it won't work if a module will be loaded 
dynamically.

If you think you need init_master - you may want to reconsider 
what you want to do and how to do it properly in nginx.

-- 
Maxim Dounin
http://nginx.org/