On Fri, Jun 15, 2001 at 04:20:33PM +0200, Sheldon Hearn wrote:
>
>
> On Fri, 15 Jun 2001 14:35:33 +0100, Oliver Cook wrote:
>
> > Just a thought, ought hupping an exim child cause it to
> > fail like it did?
>
> >From a pure UNIX point of view, the child can't really "re-read its
> configuration and start up as if new" because it's in the middle of a
> delivery. Whether it should have logged a useful message (.e.g "exiting
> on signal 1") is a different story.
It seemed a little odd behaviour to me... if a child isn't meant to
reread it's configuration on a HUP then the least it could do is ignore
a HUP rather than fail. I expect Philip has good reason for doing what he's
doing though! :)
> Regardless, this is an experience you can learn something from. :-)
> It's very good practice to reserve killall for those situations where
> you have a desperate resource problem on your system. The utility is
> horribly abused. Daemons that spawn children write out pidfiles for a
> jolly good reason. Welcome to it. :-)
Hmm.. hardly a welcome! I'm sure most of the people on this list
are well aware of where killall and pidfiles fit into the scheme
of things. :)
Ollie
--
Oliver Cook Systems Administrator, ClaraNET
ollie@??? 020 7903 3065
