Re: [spr28646] Resetting an Allegroserve worker process

At 01:19 PM 3/1/2004, John Foderaro wrote:

>  It's been my experience that browsers are very persistent and if
>their request isn't met with a valid response (e.g. the tcp
>connection just shuts down) then the browser will retry
>the request.
>
>  Try your experiment again and this time kill off the browser
>or send the browser to another page so it's not going to keep
>retrying the request.  Then see if the lisp process mysteriously
>restarts the computation.

Indeed.  It does not.  At least for the one experiment I did:

Windows, Mozilla, localhost.

Thanks!  At least that's an explanation that makes some sense.
I was beginning to be convinced that there were real demons on
my machine.

It never occurred to me that the observed behavior was being caused by
the client.


>  Also, I'd suggest that you process-reset and never process-kill
>a worker thread.

Kill enough of them and they might start unionizing.

(-:

Yes, reset seems the way to go.  Especially as there doesn't
seem to be any way to tell the Allegroserve process at startup
to replace its worker threads if they happen to die, or to tell
the Allegroserve process to make more threads.