Re: A mystery about calling a program from CGIDEV2

[Scott Klement <web400@xxxxxxxxxxxxxxxx>]

Hi Pete,

I just changed CHKSVRR4 myself, just a "quick-fix" to use non-blocking 
sockets instead of signals.

Can you try it and see if it resolves your issue?  Let me know.

You can get the code here:
http://www.scottklement.com/chksvr/ChkSvr.zip




On 1/25/2012 1:13 PM, Pete Helgren wrote:
> Yeah, that is what I suspect although Nathan's Apache Directive idea
> might have some merit.
>
> I remember reading some articles comparing the two "timeout" approaches
> (probably your articles) and I remember looking at the examples and
> using the signals/alarm approach because I could understand it and it
> seemed simpler to me.
>
> Any chance you have a functional link to an example?  Some of the
> earlier articles you did (like back in 2005) have non-functional links.
> A link to an example might get me going again.
>
> Pete Helgren
> Value Added Software, Inc
> www.petesworkshop.com
> GIAC Secure Software Programmer-Java
>
>
> On 1/25/2012 11:48 AM, Scott Klement wrote:
> > Hi Pete,
> >
> > The CHKSVR command relies on signal processing to "time-out" the
> > connection attempt.
> >
> > It's possible that Apache either disables signals in it's CGI jobs, or
> > has them handled in a special way (for example, having the submitting
> > job handle the signals for the child job.)  This sounds like it would
> > cause the problems you describe, right?
> >
> > Rewriting CHKSVR with nonblocking sockets (instead of signals) would
> > solve this problem.
> >
> > -SK