[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [patch] kill process after a timeout



On Mon, Jan 30, 2006 at 10:29:25AM +1100, Sam Lawrance wrote:
> 
> On 30/01/2006, at 1:15 AM, Florent Thoumie wrote:
> 
> >On Sunday 29 January 2006 14:38, Sam Lawrance wrote:
> >>http://people.freebsd.org/~lawrance/patch-rc.subr
> >>
> >>This patch adds the ${name}_stop_timeout variable.  When set, rather
> >>than simply wait on the PIDS after sending a -TERM signal, they will
> >>be kill -9'ed after the specified timeout in seconds.
> >>
> >>For example, with a tomcat script I'm working on, I set
> >>jakarta_tomcat41_stop_timeout=10, then
> >>
> >># sh tomcat41.sh forcestop
> >>Stopping jakarta_tomcat41.
> >>Waiting (max 10 secs) for PIDS: 42864, 42864, 42864, 42864, 42864.
> >>
> >>I need to do something like this anyway with the tomcat rc scripts, I
> >>figure it might be a useful addition to rc.subr.
> >>
> >>Thoughts?
> >
> >Looks good to me. But I wondered if we really have cases where we  
> >don't want
> >to wait with some timeout. I guess I would have modified  
> >wait_for_pid to take
> >$name_stop_timeout as a new argument (defaulting to 10, for  
> >example) instead
> >of copying wait_for_pid to a new function.
> >
> >My 0.02$ :)
> 
> We can't add a new optional argument without changing all the calls  
> to it anyway, because it takes a variable of PIDs as arguments:
> 
> wait_for_pids pid [pid ...]
> 
> I thought it was better to create a new function, just in case there  
> are other scripts and ports that use wait_for_pids.

I wonder why we haven't begun using POSIX getopts in rc.subr functions
yet.  Some of them are asking for it.  In this case:

	wait_for_pids [-t timeout] pid ...

Pretty neat, huh? :-)

-- 
Yar