Re: exec and detach a subprocess

[prev] [thread] [next] [Date index for 2005/02/22]

Igor wrote:
[...]
> Without the additional fork in "exit if fork() > 0;", background
> scripts which are started from exec(...) are killed when request is
> complete. Without the additional fork(), the PID for the bg process
> will be the same as for the "exec_helper.pl" script (since "exec" is
> used). Apache (apr?) knows this PID and it looks like it sends it a
> SIGKILL signal at request cleanup phase. I was not able trap other
> signals.

Actually we (mp2) have an almost complete control over it. The 
spawn_proc_prog was at some point modeled after mod_cgi. How about this patch:

Index: xs/Apache/SubProcess/Apache__SubProcess.h
===================================================================
--- xs/Apache/SubProcess/Apache__SubProcess.h   (revision 154417)
+++ xs/Apache/SubProcess/Apache__SubProcess.h   (working copy)
@@ -90,7 +90,7 @@
          return rc;
      }

-    apr_pool_note_subprocess(p, procnew, APR_KILL_AFTER_TIMEOUT);
+    /*apr_pool_note_subprocess(p, procnew, APR_KILL_AFTER_TIMEOUT);*/

      if (!(*script_in = procnew->in)) {
          Perl_croak(aTHX_ "broken program-in stream");

Normally it's a goodness for processes that need to run while Apache is 
running. However in your case another fork is a waste, so if the above 
works, we should call it unless an argument was passed not to call it.

        -- 
        __________________________________________________________________
Stas Bekman            JAm_pH ------> Just Another mod_perl Hacker
http://stason.org/     mod_perl Guide ---> http://perl.apache.org
mailto:stas@xxxxxx.xxx http://use.perl.org http://apacheweek.com
http://modperlbook.org http://apache.org   http://ticketmaster.com