Re: [Bug 53555] Scoreboard full error with event/ssl

[Stefan Fritsch <sf...@sfritsch.de>]

Moving the discussion to dev@

On Monday 05 October 2015 22:40:15, bugzilla@apache.org wrote:
> https://bz.apache.org/bugzilla/show_bug.cgi?id=53555
> 
> --- Comment #25 from Yann Ylavic <yl...@gmail.com> ---
> (In reply to Stefan Fritsch from comment #24)
> 
> > (In reply to Yann Ylavic from comment #23)
> > 
> > > How about not setting SERVER_GRACEFUL in close_listeners() and
> > > worker_thread()?
> > > The old generation's state could be relevent, since the new
> > > generation does not "steal" the scoreboard now (until the old
> > > worker exits).> 
> > Yes, that would proabaly be better, I'll have to test that. But it
> > would not fix the incompleteness I was referring to: The old and
> > the new process have only one process slot in the scoreboard,
> > which makes the async overview table show sometimes the info from
> > the old and sometimes from the new process, depending on who
> > updated it last.
> 
> It seems to me that the new generation's worker threads are not
> started now unless their scoreboard slot is marked SERVER_DEAD (was
> also SERVER_GRACEFUL before attachment 33158).
> So AIUI, there shouldn't be two workers using the same slot.

There are worker slots (one for each thread) and process slots (one 
for each process). You are right about the worker slots. But the 
process slot is immediately re-used by the new process. It contains 
summary information like the number of open/async/keepalive 
connections.

BTW, there is quite a bit of code in mpm event that deals with this 
taking over of some of the worker slots of an old process before it 
has completely terminated. It seems to have been a part of the 
original design inherited from mpm worker. Does anyone know if this 
feature has been intentionally disabled in mpm event?

Re: [Bug 53555] Scoreboard full error with event/ssl

[Yann Ylavic <yl...@gmail.com>]

On Tue, Oct 6, 2015 at 12:59 AM, Stefan Fritsch <sf...@sfritsch.de> wrote:
>
> There are worker slots (one for each thread) and process slots (one
> for each process). You are right about the worker slots. But the
> process slot is immediately re-used by the new process. It contains
> summary information like the number of open/async/keepalive
> connections.

Hm, you are right.

>
> BTW, there is quite a bit of code in mpm event that deals with this
> taking over of some of the worker slots of an old process before it
> has completely terminated. It seems to have been a part of the
> original design inherited from mpm worker. Does anyone know if this
> feature has been intentionally disabled in mpm event?

Dunno, but maybe we could use atomic operations for the process
scores, so that all the generations can report simultaneously...