Re: FW: Re: [Exim] Exim on a single-user system

Page principale
Supprimer ce message
Répondre à ce message
Auteur: Vadim Vygonets
Date:  
À: exim-users
Sujet: Re: FW: Re: [Exim] Exim on a single-user system
Quoth Matthew Byng-Maddick on Wed, Jan 02, 2002:
> On Wed, Jan 02, 2002 at 09:45:56AM -0500, Derek Broughton wrote:
> > Right. Similarly fetchmail prefers to deliver to port 25 than to
> > sendmail. I wonder if it's an assumption that the SMTP server is a
> > standalone daemon and therefore has no startup overhead? Of course, not
> > necessarily a valid assumption.


Hmm.

: narn:~%4; size /SBIN/exim
text    data    bss     dec     hex
415874  14300   7040    437214  6abde   /SBIN/exim


I'd say that loading 14K of data[0] off the disk doesn't take
much more time than making DNS lookups and whatnot.

> Fetchmail is a pile of crap, because it often loses mail, due to this. It
> ignores any error return codes in delivering the message, but happily
> deletes it from the pop/imap server anyway.


Agreed, and I've seen it happen. Mail should NEVER be lost.
Unless it's an undeliverable bounce message, or is explicitly
blackholed, that is. What I mean is: mail from a real human to a
real human should NEVER be lost.

I heard that getmail is a good program that does things right.

> This is a good reason for NOT using SMTP delivery, IMO.


Or implementing it correctly, including some sort of queue. Mutt
(which I use), when encountering a problem running sendmail,
where you can retry, postpone or abort the message. It also
shows you the full error message.

Vadik.

[0] If an Exim is already in core, text is in core too.  If it's
    swapped out, loading text from disk will happen whether Exim
    is run or woken up from the network.  Basically.


--
Of course [nobody reads the docs that come with the OS] -- that
would be too easy and too quick.  People know that the Unix Way
is difficult and they prefer to keep it that way.
        -- Greg Black