[ale] Boot script becomes a zombie?

[Michael H. Warfield]

On Mon, 2010-02-22 at 15:06 -0500, Brandon Checketts wrote: 
> Conceptually, you should be able to send a SIGCHLD signal to the parent process
> to tell it to clean up (ie: call wait()) after its children.  In reality, I have
> no idea if that actually works.

No, it doesn't work in all cases.  The problem I have run into is that
certain things (like sshd and possible cron) will wait on the open file
descriptors to close.  Even if they get a SIGCHLD signal, if the file
descriptors are still open, the logic is that they still might have
something to do.  So, if you start a child process, that child process
inherits all of your file descriptors.  Even if you exit, those pipes
are still being held open by the (grand) child.  If the parent (of your
process) is waiting on those to close, it will not get it while its
grandchild holds them open.

> kill -15 {parent PID}
> 
> Thanks,
> Brandon Checketts
> 
> 
> 
> Lightner, Jeff wrote:
> > In that situation sometimes (though not always) killing the parent will
> > get rid of the zombie.
> > 
> > As an HP presenter used to say at the start of his performance
> > presentations:
> > You can't kill a zombie because IT'S ALREADY DEAD!
> > 
> > -----Original Message-----
> > From: ale-bounces at ale.org [mailto:ale-bounces at ale.org] On Behalf Of
> > Michael H. Warfield
> > Sent: Monday, February 22, 2010 2:29 PM
> > To: Atlanta Linux Enthusiasts - Yes! We run Linux!
> > Cc: mhw at wittsend.com
> > Subject: Re: [ale] Boot script becomes a zombie?
> > 
> > On Sun, 2010-02-21 at 23:05 -0500, James Sumners wrote: 
> >> I have a script that looks runs several other scripts in turn. I use
> >> this script to startup several FastCGI processes. This is the script:
> > 
> >> ##########
> >> #!/bin/bash
> >> find /var/www/fastcgi/scripts/ -type f -execdir '{}' 2>&1 1>/dev/null
> > \;
> >> return $?
> >> ##########
> > 
> >> I then have the following crontab entry for root:
> >> @reboot /path/to/boot_cgi_bin.sh
> > 
> >> Every time I reboot my server, this boot_cgi_bin.sh script turns into
> >> a zombie (the cgi bins do get started). Does anyone see a reason why
> >> this script would hang up as a zombie process?
> > 
> > A zombie process means that process terminated but the process (which
> > would seem to be cron in this case) which started it has not reaped the
> > child's status.  It's really all gone and only exists as an entry in the
> > proc table.  This CAN happen if a child it invokes leaves it's file
> > descriptors open and they're pipes back to the parent (it's what causes
> > ssh to refuse to terminate after running a script which restarts
> > services.).  I had to deal with that in the smbmount program when I was
> > maintaining it.
> > 
> > Do a ps -lp {process id} on that process id and then see what the parent
> > process is (may be a shell).
> > 
> > Might also redirect stdin from /dev/null just in case one of those
> > scripts you're invoking is doing something funny with stdin.
> > 
> > Mike
> _______________________________________________
> Ale mailing list
> Ale at ale.org
> http://mail.ale.org/mailman/listinfo/ale
> See JOBS, ANNOUNCE and SCHOOLS lists at
> http://mail.ale.org/mailman/listinfo
> 

-- 
Michael H. Warfield (AI4NB) | (770) 985-6132 |  mhw at WittsEnd.com
   /\/\|=mhw=|\/\/          | (678) 463-0932 |  http://www.wittsend.com/mhw/
   NIC whois: MHW9          | An optimist believes we live in the best of all
 PGP Key: 0x674627FF        | possible worlds.  A pessimist is sure of it!
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 482 bytes
Desc: This is a digitally signed message part
Url : http://mail.ale.org/pipermail/ale/attachments/20100222/2a7b5398/attachment.bin