Additional recipients within same lose mail when quarantined/released...

[Glenn Steen]

On 15/03/07, am.lists <am.lists at gmail.com> wrote:
> On 3/15/07, Glenn Steen <glenn.steen at gmail.com> wrote:
> > On 15/03/07, Glenn Steen <glenn.steen at gmail.com> wrote:
> > > On 15/03/07, am.lists <am.lists at gmail.com> wrote:
> > > > On 3/15/07, Glenn Steen <glenn.steen at gmail.com> wrote:
> > > > > On 15/03/07, am.lists <am.lists at gmail.com> wrote:
> > > > > > Sorry for the ramble of a subject line, but what I'm seeing is
> > > > > > essentially lost mail to additional recipients.
> > > > > >
> > > > > > Scenario: (Config: Postfix, Mailscanner, Mailwatch.)
> > > > > >
> > > > > > Bob sends John and Mary an email. It gets tagged as spam. Both are at
> > > > > > somedomain.net. As John's listed as the first to address in the
> > > > > > envelope, he gets the message in his quarantine. He releases it. Mary
> > > > > > knows nothing of the email. No quarantine file for her (in MailWatch),
> > > > > > and when John releases it, it only goes to John.
> > > > > >
> > > > > > Variation:
> > > > > >
> > > > > > Bob sends to Tom and John. John is at somedomain.net but Tom is at
> > > > > > anotherdomain.us. Both are handled by my mailscanner. Same exact
> > > > > > behavior as above.
> > > > > >
> > > > > > I've been googling this and the solutions I've found have been really
> > > > > > ancient (2003 vintage), and as such, several versions of released code
> > > > > > old.
> > > > > >
> > > > > > I'm wondering if there's some new way of fixing this issue? I don't
> > > > > > particularly want to rig up a second instance of postfix on some odd
> > > > > > port just to fix this... unless I have to. 8-(
> > > > > >
> > > > > > Angelo
> > > > > Not yet, there isn't... Unless you patch MailWatch to handle this
> > > > > (ISTR there being some patches for this floating around).
> > > >
> > > >
> > > > Well, I've made a right mess of my kit by trying Joost's referred
> > > > article. There's at least one or two material typos in it, and it's a
> > > > dickens to troubleshoot. I'm meddling a bit more with it before
> > > > reverting back.
> > > >
> > > > Angelo
> > > Being the author of that article, I'd like to be alerted to what typos
> > > you find... Might help the next guy/gal:-).
> >
> > Right, found one I thought I'd fixed (postfix.in/transport and
> > postfix/transport.in mixup). Fixed that one good this time. Please
> > reload your pages:-).
> >
> > Was there anything else (apart from a nonsensical "compare this OR
> > this" example near the end:)?
> >
>
> Well, for one, the /etc/postfix/transport.in =>
> /etc/postfix.in/transport was one problem.
Check. That one will make life very hard if one missed it, yes. As
said ... all fixed now:).

> Also, let's consider why I'm after this article:  the original post:
> about releasing properly to multiple recipients from within MailWatch.
> It was confusing to me that if I went out and grabbed mini_sendmail
> that I wouldn't need the postfix instance on port 10027. But it seems
> that the mini_sendmail is just a remailer, telling it to connect to
> localhost:10027... which implies that the 10027 instance /does/ need
> to be there.
I'm not sure I follow you here... It should be fairly obvious that
that part of it was written at two different times, the first when I
was just theorising about necessary tools to overcome the whole "how
do I release mail when I no longer can whitelist 127.0.0.1" problem,
the second a more thorough job on describing how to go about it... I'm
fairly certain I never imply what you're saying above. But then again,
the wording is a bit messy:-).


> And... I'm not sure if it was needed or not, but I created
> /etc/init.d/postfix.in - copied from /etc/init.d/postfix, adding -c
> /etc/postfix.in as the config files to use. It wasn't clear if this
> was needed or not.
Why would you do a silly thing like that? Didn't you read step #13? It
rather clearly (IMO:-) states that the MailScanner init script will
find and use both instances.

> I run as a gateway, and as such, rely on whitelisting 127.0.0.1 for
> releasing. The doc says to remove whitelisting of 127.0.0.1. But then
> what? I understand that it needs to be removed to pass the mail from
> one spool to the other, but how do I then resend it back out without
> the double (or looping) scanning?

The whole ending note is _only_ about solving this problem.... After
the "And the time is now:-)"... Did you read that part carefully?

> As of now, I've executed a rollback. But I'm missing something I
> think. If you have any ideas on what I've not done correctly, I'd love
> to hear.
>
> Backout:
>
> Restored /etc/postfix-bak to /etc/postfix  (I copied it to -bak before
> I started mucking)
> Restored /var/www/html/mailscanner-bak to /var/www/html/mailscanner
> Removed /etc/init.d/postfix.in
>
> To be sure, I init-6'ed. Now, after the reboot, no mail is coming
> through, everything appears to be going into the
> /var/spool/postfix.in/hold directory. (I know this based on the reboot
> being at 15:09 and files in /var/spool/postfix.in/hold having a 15:18
> timestamp.

Did you muck about further with the init scripts? That might be something...
I'm off to my commuter train, but ... keep me posted, I'll take up the
thread once I'm home.

Cheers
-- 
-- Glenn
email: glenn < dot > steen < at > gmail < dot > com
work: glenn < dot > steen < at > ap1 < dot > se