Subject: Re: sendmail configuration
To: Greg Hudson , Valtteri Vuorikoski <email@example.com>
From: Ask Dr. Stupid <greywolf@defender.VAS.viewlogic.com>
Date: 11/02/1995 14:16:41
#define AUTHOR "ghudson@MIT.EDU (Greg Hudson)"
* > Shouldn't sendmail do this if it's running as sendmail -bd -q30m ?
* > (try to de-queue mails every 30 minutes)
* Yes, but sendmail isn't turned on in the default /etc/netstart. At
* MIT, we don't turn on sendmail on our client machines, since all
* incoming mail comes through the mail hubs.
If you spool incoming AND outgoing mail to the mail hubs directly, then
that makes sense, but even if you're going to be sending mail from
a machine, you should start sendmail on it.
It's kind of a pity you can't just do a 'sendmail -send-only' and have
it just sit there processing the queue as needed, and have it refuse in-
coming connections (if, as you say, everything gets routed through the
mail hubs, and all the aliases and translations are set correctly). If
configured correctly, this would permit the client to send mail outward
and not require it to listen for incoming mail (since even if it were
to bounce back, the mail hub would properly re-route it).
I can't see a single valid reason not to run sendmail at all. Even if
you're the only user on a standalone machine, some processes want to send
you mail if they blow up. Granted, it's configurable, but there's still
a good chance that running sendmail is a desirable thing.
* It does seem like a shame to have such a cron job when many mail users
* are just going to turn on the sendmail daemon, but the current
* situation seems patently incorrect.
Didn't you just say the same thing twice here?
#undef AUTHOR /* "ghudson@MIT.EDU (Greg Hudson)" */