Ahh, I see what you mean. Unfortunatly this does not work on traditional
BSD (according to google). The zombies are still created.
Here is a patch for FreeBSD.
--- pcpd.c Sat Mar 1 01:19:08 2003
+++ pcpd.c.patched Sat Mar 1 01:23:45 2003
@@ -2060,6 +2060,7 @@
static void start()
{
+ struct sigaction sa;
int lockfd=ll_daemon_start(LOCKFILE);
int pubsock;
int privsock;
@@ -2082,7 +2083,11 @@
ll_daemon_resetio();
ll_daemon_started(PIDFILE, lockfd);
- signal(SIGCHLD, SIG_IGN);
+
+ memset(&sa, 0, sizeof(sa));
+ sa.sa_handler=SIG_IGN;
+ sa.sa_flags=SA_NOCLDWAIT;
+ sigaction(SIGCHLD, &sa, NULL);
for (;;)
{
Thanks for the help.
Ben
----- Original Message -----
From: "Sam Varshavchik" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Sent: Friday, February 28, 2003 11:42 PM
Subject: [sqwebmail] Re: zombie processes with pcpd
> Ben Pfountz writes:
>
> > Hello list,
> >
> > I am using sqwebmail with calendaring in groupware mode. My OS is
FreeBSD
> > 4.7-STABLE. It works really well, except that I have many pcpd zombie
> > processes that seem to collect as I browse through my callendar. A
quick
> > scan through pcpd.c shows a call to fork when the pcpd server receives a
> > connection, but I cannot find a place where it then waits for the forked
> > process. Am I missing something? Is this done on purpose?
>
> SIGCHLD should be set to SIG_IGN:
>
> signal(SIGCHLD, SIG_IGN);
>
> This should prevent accumulation of zombie processes.
>
>
>
>
>
-----
Ben Pfountz
Computer Science Undergraduate
Virginia Tech
Computer Systems Engineer
Center for Power Electronic Systems
Virginia Tech
