Re: Wheel::Run and Proc::Daemon troubles

On Sep 1, 2005, at 15:56, Hicks, Matthew wrote:

> I'm writing a program that can run interactively, or as a daemon.  It
> spawns a few sessions with POE::Wheel::Run that work fine  
> interactively.
> If I call Proc::Daemon::Init, the wheels generate a single error
> signal--operation: read, errnum: 0, errstr: <blank>--then a .
Read error 0 is EOF.

> Is this a side-effect in Wheel::Run due to Proc::Daemon closing the
> initial process' filehandles?  If so, what is the 'proper' way to
> daemonize a POE program in order to support Wheel::Run?
>
> I daemonize before calling kernel->run().  STDERR is re-opened to a
> logfile in order to capture trace output, but I can't find anything  
> that
> looks like a meaningful error message.
_start handlers may be executed before POE::Kernel->run().  If the  
wheels are created there, Proc::Daemon may be interfering with them.
Can you daemonize the program before starting any sessions?

> Any pointers how to continue debugging/solving this issue?
Determine what Proc::Daemon::Init is doing.

How would that force child processes to close their sockets, or  
simulate the behavior from the parent process' point of view?
--
Rocco Caputo - rcaputo@xxxxxxxxx