LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Re: Ldirectord Redhat EL3 SSL checking problem

To: "LinuxVirtualServer.org users mailing list." <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: Ldirectord Redhat EL3 SSL checking problem
From: Horms <horms@xxxxxxxxxxxx>
Date: Mon, 8 Nov 2004 17:41:28 +0900
On Fri, Nov 05, 2004 at 02:42:16PM -0800, Seth Daniel wrote:
> On Fri, Nov 05, 2004 at 11:05:25PM +0100, Johan Elmerfjord wrote:
> > Hi Phil,
> > 
> > It looks very much like the problem I have.
> > I addressed this issue in the Linux-HA mailinglist, but have not recievied
> > any suggestions yet.
> > I found a 'solution' that works for me. But I have no idea of what other
> > impacts it may have.
> > 
> > See my problem and 'solution' in my original post:
> > http://lists.linux-ha.org/pipermail/linux-ha/2004-November/012771.html
> 
> This is very similar to the solution I arrived at when I started to see
> this problem.  The difference being that I set $SIG{'CHLD'} to undef
> instead of commenting out the line (local $SIG{'CHLD'} = undef;).
> 
> The perlipc perldoc page suggests that if you set $SIG{'CHLD'} to
> 'IGNORE' then wait() will return -1.  It would also appear to be setting
> $? to -1 as well which is where the problem comes in.  There isn't
> really a reason to set $SIG{"CHLD"} to IGNORE since wait() is always
> called.  

Ok that makes some sense. I would guess that slightly older
versions of perl don't exhibit this behaviour. Did setting
$SIG{'CHLD'} to undef work? What I really want is for
there to be no handler on SIGCHLD and for the retun value
to be picked up by wait. It seems my assumtion that 
$SIG{'CHLD'} = "IGNORE" would do that just because that is
the way it works in C is wrong.

-- 
Horms
<Prev in Thread] Current Thread [Next in Thread>