Re: Problem with shellcommand and background job

[Christian Pearce]

Here is the exact code from our SysNav 5.0 release.  This is what I use
to restart the clients and the server if need be.  Notice I renamed my
runbb.sh to runbb-server.sh for our proxynodes.

processes:
  
    _proxynode.bb_installed::
        "runbb-server.sh"
            restart "$(bb_prefix)/bb/runbb-server.sh start"
            owner=$(sysnav_user)
 
    bb_installed::
        "bb-local"
            restart "$(bb_prefix)/bbc/runbb.sh start"
            owner=$(sysnav_user)

Here is an example of something that held onto the STDOUT, and STDERR.

    # Reboot if scheduled automation occurs
    reboot::
        '$(shutdown) -y -g 300 -i 6 "SysNav is rebooting the server to
install patch Clusters." > /dev/null 2>&1 &'
            useshell=true

Notice I pipe both STDOUT and STDERR to /dev/null and then background
it.  This works like a champ.

There is also this like in the wiki.

http://cfwiki.org/cfwiki/index.php/Hung_Shell_Command

On Thu, 2005-02-17 at 15:30, David Douthitt wrote:
> Manon Goo wrote:
> > I  have tried to restart BigBrother from cfengine (2.1.13 )  the 
> > BigBrother restart script starts some other processes in background
> > but these to not close the STDIN and STDERR. Now cfengine habs
> > infinite unless I redirect the output of The restart script.
> 
> I don't know if there is a way around it, but I believe that is expected 
> behavior, though incorrect behavior.  The programs should close STDIN 
> and STDERR, but if they do not then things hang.
> 
> I ran into this trying to get some programs to run during system startup 
> in initialization scripts.
> 
> The real question is: can cfengine work around this?  I believe you can 
> use a shell script to "hold" the process which doesn't close STDIN and 
> STDERR properly.
-- 
Christian Pearce
http://www.commnav.com
http://www.perfectorder.com