At 01:47 PM 6/29/2004, Mathihalli, Madhusudan wrote:
>: -----Original Message----- >: From: Colm MacCarthaigh [mailto:[EMAIL PROTECTED] >[SNIP] >: > Upon doing a "apachectl stop", shouldn't the CGI processes >: (forked by >: > mod_cgid or mod_cgi) also exit ? >: >: It depends, if they call setsid() and so on - there's no >: particular reason they should. Ordinarily - yes. > >In case of setsid(), I have no idea what the behaviour should be :-). > >For the 'Ordinarily' case - it doesn't do it today. It can be easily >done by having a "kill (0, SIGTERM)" just before mod_cgid/apache exits - >but I don't know if it's portable. Wouldn't a SIGHUP make more sense? Some CGI's may override, seeing as they were invoked to perform an operation (non-abortable) as opposed to retrieving results. The client does HUP in those cases, from a coder's point of view.
