Kill signal testing of various mpms...

["Paul J. Reder" <re...@raleigh.ibm.com>]

I have done some testing sending various kill signals to various MPMs.
The results are summarized at http://24.25.12.102 (follow the link to
the kill test results).

Basically the worker MPM had difficulties, and threaded had a hard time
doing graceless shutdown/restart while under a request load. But all MPMs
did what they are supposed to when no request load was present. So Greg is
correct in saying that the threaded MPM does shutdown/restart immediately
under no load.

Please let me know if you have different results than what is published.
I also only have results for unix MPMs run on Linux. I don't have a setup
to test OS/2 or Windows.

I will work to keep these results up to date as changes occur.

-- 
Paul J. Reder
-----------------------------------------------------------
"The strength of the Constitution lies entirely in the determination of each
citizen to defend it.  Only if every single citizen feels duty bound to do
his share in this defense are the constitutional rights secure."
-- Albert Einstein

Re: Kill signal testing of various mpms...

[Greg Ames <gr...@remulak.net>]

"Paul J. Reder" wrote:
> 
> I have done some testing sending various kill signals to various MPMs.
> The results are summarized at http://24.25.12.102 (follow the link to
> the kill test results).
> 
> Basically the worker MPM had difficulties, and threaded had a hard time
> doing graceless shutdown/restart while under a request load. 

yeah, none of the threaded MPM's have logic to blow active requests out
of the water AFAIK. For shutdown, threaded should be sending SIGTERM,
then SIGKILL to the child processes and putting messages in the log once
it gets impatient, and that should do the job.

I've heard an intriguing suggestion for speeding up non-graceful, but
let's get a stable release out the door before we start tinkering too
much.  ssi's aren't working reliably at the moment.

Greg