httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jeff Trawick <traw...@attglobal.net>
Subject Re: cvs commit: httpd-2.0/server/mpm/worker worker.c
Date Wed, 20 Mar 2002 15:51:50 GMT
Greg Ames <gregames@remulak.net> writes:

> trawick@apache.org wrote:
> 
> >   +/* XXX unfortunate issue:
> >   + *     with apachectl restart (not graceful), previous generation dies
> >   + *     abruptly with no chance to clean up scoreboard entries; when new
> >   + *     generation is started, processes can loop forever in start_threads()
> >   + *     waiting for scoreboard entries for threads of prior generation to
> >   + *     get cleaned up...  
> 
> yikes!  
> 
> Did that break when the scoreboard moved to the process pool?  Looks like
> ap_cleanup_scoreboard won't run any more.

no, it is unrelated...  it was occurring before yesterday's commits,
probably for a long time (I ran into the annoying series of
SIGTERM+SIGTERM+SIGTERM+SIGKILL messages when testing some of my
changes and went back and then verified that it happened on cvs head)

but some knowledge of how the scoreboard used to work shows that the
comment shouldn't be right

Here is the actual sequence of events that I did:

  start
  (wait)
  graceful
  (wait)
  restart
  (wait)
  stop

This wasn't a once or twice thing.  Always after this sequence I saw
the annoying messages when I did apachectl stop.  Sometimes (40%?) I
saw a few such messages also when I did apachectl restart.  This was
on multiple OSs as well.

I only noticed the problem (new children with no place to go) when I
did the "restart."  But a restart clobbers (well, *did*) the
scoreboard so how can a newly-started child be waiting forever for the
scoreboard slots in this scenario?

While the problem really exists (children that never find any slots),
I don't honestly know the cause and intend to remove most of the
comment so that it doesn't lead anybody down the wrong path.

Also, I think I'll put in a debug message every 180 iterations (3
minutes) if the child process hasn't made any progress taking over
more slots.

-- 
Jeff Trawick | trawick@attglobal.net
Born in Roswell... married an alien...

Mime
View raw message