The penny just dropped, if I do:
pidof php5 data.php
the PID is returned.
Looks like the initial problem ... ie from the directory that data.php is in,
that it couldn't be invoked by ./data.php but worked with php5 data.php seems
to influence how pidof now works.
Well, looks like I have a work-around.
Thanks,
Dave
> Date: Mon, 2 Mar 2015 10:56:48 +0100
> From: m...@pengutronix.de
> To: j...@pengutronix.de; oselas@community.pengutronix.de
> CC: dave_fest...@hotmail.com
> Subject: Re: [oselas] Latest mini2440BSP bash pidof problems
>
> On 03/02/2015 10:54 AM, Juergen Borleis wrote:
> > On Monday 02 March 2015 10:38:42 Marc Kleine-Budde wrote:
> >> On 03/02/2015 10:30 AM, Juergen Borleis wrote:
> >>> Maybe you should do more validating tests.
> >>>
> >>> PROC_TO_KILL=`pidof data.hhp`
> >>> if [ -n "${PROC_TO_KILL}" ]; then
> >>> ps "${PROC_TO_KILL}" > /dev/null
> >>> if [ $? -eq 0 ]; then
> >>> kill "${PROC_TO_KILL}"
> >>> fi
> >>> fi
> >>
> >> This is all racy,
> >
> > Sure. But better than not to check.
> >
> >> what about pkill?
> >
> > Killing processes via pattern sounds like a horror to me. The correct
> > solution
>
> What's the difference to pidof? pkill and pgrep have even more ways to
> specify the pattern.
>
> > would be to make the "data.php" provides its own PID file to kill the
> > correct
> > process instead of guessing it. Just an idea.
>
> ACK, or use systemd.
>
> Marc
>
> --
> Pengutronix e.K. | Marc Kleine-Budde |
> Industrial Linux Solutions | Phone: +49-231-2826-924 |
> Vertretung West/Dortmund | Fax: +49-5121-206917-5555 |
> Amtsgericht Hildesheim, HRA 2686 | http://www.pengutronix.de |
>
