NetNews Usenet Archive 1992 #30/NN_1992_30.iso/spool/comp/unix/sysv386/17350

Internet Message Format | 1992-12-12 | 2.3 KB

Xref: sparky comp.unix.sysv386:17350 comp.mail.sendmail:2976
Newsgroups: comp.unix.sysv386,comp.mail.sendmail
Path: sparky!uunet!pipex!demon!orbital.demon.co.uk!steveh
From: [e-mail] (Stephen Hebditch)
Subject: Re: DELL DOES NOT SUPPORT FULL IMPLENTATON OF UNIX SVR4; SENDMAIL UNSUPPORTED
Message-ID: <[e-mail]>
Organization: TQM Communications, London, UK
References: <[e-mail]> <[e-mail]> <[e-mail]>
Date: Fri, 11 Dec 1992 14:32:31 GMT
Lines: 30

In article <[e-mail]> [e-mail] (bill davidsen) writes:
>In article <[e-mail]>, [e-mail] (Stephen Hebditch) writes:
>| The Sun 4.1 version that came with Dell seems to lose the occasional
>| incoming SMTP message. IDA Sendmail is much nicer anyway, and
>| relatively straightforward to get going under SVR4. I was also able to
>| patch it so that it worked properly with a standalone machine that
>| makes occasional connections to send e-mail back over slip. The problem
>| is that with no local network up, attempting to connect to a remote
>| address returns EADDRNOTAVAIL: Cannot assign requested address.
>| Sendmail will treat this as a permanent rather than a temporary error,
>| so sending mail to any address which is to go out via SMTP will result
>| in it being returned immediately. Making this a temporary error, like
>| the ETIMEDOUT or EHOSTUNREACH you would get instead if you did have a
>| network running, means items then wait in mqueue until you connect the
>| slip link and flush the queue.
>
>  I'm not a sendmail guru, but I think you have a setup problem here.
>What you want is to have sendmail queue the mail and check the queue
>every N minutes to send stuff out. This is not exactly a bug in
>sendmail, since it's documented to work that way. I *think* you can do
>it on a site basis.

Using the 'expensive mailer' option I think you mean, whereby mail
destined for a particular mailer is left in the queue until you
explicitly flush it. I hadn't actually considered that but I guess it
would work. I feel though that having EADDRNOTAVAIL as a temporary
error is probably more correct anyway.

--
Stephen Hebditch  TQM Communications  [e-mail] [phone]