httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Roy T. Fielding" <field...@liege.ICS.UCI.EDU>
Subject Re: feature idea
Date Mon, 10 Jun 1996 23:28:47 GMT
>   I think one might accomplish this by fielding/catching a different
>   signal that HUP,
> 
> Ummm... here's what's happening.  When the "caretaker" server process
> (the one whose pid is in the pidfile) receives a SIGHUP, it does a 
> killpg(), again with a SIGHUP, to the process group containing all of
> the server child processes.  
> 
> Right now, on receipt of a SIGHUP, they all shrivel up and die.  What
> they *ought* to be doing is finishing whatever they are working on, and
> then going away the next time they reach the top of the loop in child_main().

What about children that are known to be stuck in a long request timeout?
Will this kill them quickly, or should we have separate signals for
kill-eventually and kill-right-now-dammit?

......Roy

Mime
View raw message