[Mimedefang] kill -HUP [MD processes]

[David F. Skoll]

On Mon, 1 Dec 2003, Brent J. Nordquist wrote:

> > 	md-mx-ctrl reread

> I've always wondered, is there any race condition there?

No.

> In other words,
> if I change mimedefang-filter but don't do anything to make the slaves
> re-read, will they re-read it on their own at some point?

Yes, eventually, as old slaves are killed off and new ones started to
replace them.

> Is there any danger of a slave reading an intermediate save of
> mimedefang-filter from the editor that isn't ready for prime-time
> (hasn't passed mimedefang.pl -test for example)?

Yes, there is.  If a slave is killed during the normal course of
operation, after processing 200 messages or whatever, and a new one
starts up while you are editing the file, then you could have a problem.
You should always create a new file (mimedefang-filter.new), test it
like this:

	mimedefang.pl -f /etc/mail/mimedefang-filter.new -test

and when you're happy, change it atomically:

	mv -f /etc/mail/mimedefang-filter.new /etc/mail/mimedefang-filter

Regards,

David.