Re: Generate Bouncemail from nonzero exit code returned by procmail

Matthias Andree via Fetchmail-users <[email protected]> Wed, 30 Oct 2024 22:22:05 +0100
Newsgroups gmane.mail.fetchmail.user
Message-ID <[email protected]>
Am 30.10.24 um 21:10 schrieb Marvin Dickhaus via Fetchmail-users:
> Hey all,
>
> I have the following setup:
> local mailbox --> fetchmail --> procmail --> another local mailbox
>
> My .procmailrc is configured to reject certain mails with exitcode 67

Procmail should not be used. Transition to something that hasn't got a
flawed design or been in disrepair for 30 years.

> My .fetchmailrc is configured not to keep mails and fetch all.
>
> My expectation would be, that mails "rejected" by procmail would be
> bounced by fetchmail.
> However that is not the case. Instead fetchmail fetches the mails on
> every call and doesn't generate any bouncemails for them.
> Is there anything I'm doing wrong, or is it just not possible to
> generate a bouncemail in this scenario?
> I've also tried setting "antispam 067" in .fetchmailrc but that did
> not change anything.

You should not bounce mail after you've downloaded ever. Either you can
reject it in the SMTP conversation before you accept it, or if you have
accepted it you delete or discard it, but you don't bounce because you
have no proof that the sender address in the e-mail is correct.

Fetchmail has a "softbounce" option that is ON by default to avoid mail
loss because too many new users used to shoot holes in their feet.

If you want to delete mail, then the recourse is adding these two lines
atop your .fetchmailrc:

set no bouncemail
set no softbounce

> Note: I did also install featchmail on Archlinux (version 6.4.39) but
> experienced the same behavior here.

Thanks for that attempt because I was almost tempted to tell you that
fetchmail 6.3.x has been unsupported for the past five years.

fetchmail 6.5.0 has been released yesterday and I don't want to spend
time on 6.4.x either any more, but I won't blame a distro that takes a
few days to adopt it, and 6.4.39 qualified because 6.5.0 is the same.

FreeBSD's "latest" ports tree has just updated.