rc functions don't allow processes to shutdown

Doug Barton dougb at FreeBSD.org
Fri Aug 31 22:49:00 PDT 2007


On Fri, 31 Aug 2007, Sean Bruno wrote:

> Doug Barton wrote:
>> Assuming that lack of an affirmative _enable variable is a constant, the 
>> only way that a service can be started is with either onestart or 
>> forcestart. The symmetry here would be to stop it the same way.
>> 
> This may be symmetrical, but I question whether or not the 'correct' behavior 
> is symmetrical.
>
> I don't see the benefit to the end user in this implementation, e.g. trying 
> to disable a running service.  If a new-ish admin edits the rc.conf prior to 
> shutting down the service, there could be some consternation.

I would chalk this up to one of the many things that an inexperienced 
admin needs to learn about system administration. I am heavily in favor of 
reasonable changes to improve usability, however I think you're tilting at 
a windmill here.

Here you have articulated one edge case where the "expected" behavior is 
something other than what happens now, but I completely fail to see how 
any benefit that might accrue from changing the code to "fix" this edge 
case overcomes the cost of the POLA violation we'd commit by changing the 
code at this late stage in the life of rc.d.

> However, my entire idea of how to shut down a process in FreeBSD may be 
> flawed, and I may be the one who is in need of a 'design change'.  :)

I don't think it's flawed, but I think you're preoccupied with something 
that is not that big of a problem that already has a convenient solution.

Doug

-- 

     This .signature sanitized for your protection



More information about the freebsd-hackers mailing list