httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Manoj Kasichainula <ma...@io.com>
Subject Re: Whacking shared memory on graceful restarts is bad
Date Sun, 30 Jul 2000 12:22:33 GMT
On Sun, Jul 30, 2000 at 04:50:08AM -0700, dean gaudet wrote:
> actually it looks like the MPM stuff itself isn't really sufficient to
> handle the case where you've got a single process server... because pconf
> itself could still be in use by a bunch of threads gracefully finishing up
> requests.
>
> it kind of only works now for servers which have a parent and a bunch of
> forked children.

I would implement any single-process MPM using a monitor process
like Windows has in 1.3. This would look like dexter with a single
child process. Any good reason not to?

If it's important to deal with the single-process case (and I have no
reason to suspect that it is ATM), then I guess the correct solution
would be to not destroy pconf until the last process/thread dies.
This feels like the right thing to do anyway.


Mime
View raw message