That is NOT a bug. Two processes do NOT fight over the same chunk of the scoreboard.
Each process that grabs a slot in the scoreboard has a thread as well as a process.
Since the
threads pay attention to what other processes are already using slots in the
scoreboard, we
don't overwrite anything.
It has been this way since July 3rd, and fixes the problems that everybody was
complaining about with
the threaded MPM.
Feel free to fix any bugs you find, but make sure they are bugs. Having two processes
using the
same slot is not a bug. It solves a bug, namely that the server stopped responding
when gracefully
restarting while serving a lot of very large requests.
We may be creating too many processes at one time, but I don't believe that we are. I
believe that the
segfaults that we are seeing have nothing to do with the scoreboard, but rather that
they are a problem
with the machine itself, which is a problem that has been reported many times on this
list.
The threaded MPM logic as working right now has been reported to work on Linux, and
Solaris, as well, I
believe, on a few other platforms.
Ryan
On Friday 27 July 2001 11:53, Greg Ames wrote:
> Ryan Bloom wrote:
> > Unless somebody objects, I will commit my latest patch to the threaded
> > MPM later this weekend. This is the patch that implements a multi-process
> > single acceptor/multi-worker model.
> >
> > This will allow us to finish fixing the threaded MPM.
>
> I object. Threaded is a mess as it stands now, with multiple new
> processes allowed to grab the same chuck of scoreboard. We need to fix
> that before destabilizing it further.
>
> Greg
--
_____________________________________________________________________________
Ryan Bloom [EMAIL PROTECTED]
Covalent Technologies [EMAIL PROTECTED]
-----------------------------------------------------------------------------
