RE: [PATCH] forcefully kill of threads in worker MPM

[Sander Striker <st...@apache.org>]

> From: gregames [mailto:gregames]On Behalf Of Greg Ames
> Sent: 29 April 2002 18:06

> Aaron Bannert wrote:
> > 
> > This ugly thing kills off threads that are running long-lived
> > requests when we want to do a graceless shutdown of the server.
> 
> +1 in concept and eyeballing the code.  Something drastic seems in order, and I
> haven't heard a better alternative.

+1 from here too.  The only alternative I see is apr_thread_cancel, and
I don't feel like getting into that discussion right now.
 
> > I tested it by running ab with concurrency of 100 against a 10MB
> > file and running "bin/apachectl stop ; ps -ef | grep httpd". I
> > get a bunch of [info] errors in the error log of this form:
> > 
> > [Sun Apr 28 17:04:26 2002] [info] (9)Bad file number: core_output_filter: writing data to the network
> > 
> > But hey, when you're killing off your server gracelessly what do
> > you expect? We might want to add an error message before all those
> > core_output_filters that is something like "We're going to prematurely
> > kill off all the current connections now..."
> 
> Doesn't the parent log a message already when it catches the signal?  If there
> are several child processes running, I don't think it would help to have all of
> them log the same message, except perhaps for debugging.  If we could change the
> log level of the messages from core_out once the child gets into this state,
> that might be better.
> 
> Greg

Sander