On Tue, Jan 29, 2008 at 03:42:00PM -0800, Jo Rhett wrote:
> On Jan 29, 2008, at 2:26 PM, Dave McGuire wrote:
> > Because 99% of the time, end-users are complete morons who couldn't
> > find their own butts with both hands, a funnel, and a roadmap. The
> > vast majority of them can't handle sorting based on tagging criteria.
> This is why I am curious. Trying to get users to log into a
> different system to check the quarantine has never yielded good
> results in my mind. Sorting it into a folder they can check on their
> mail client of choice always seems to work, they understand it, etc.
> That's why I question the value of server-side quarantine.
> > If amavisd's quarantining code is removed for any reason, I'll
> > have to write a replacement myself, because I won't run a mail
> > server without that functionality.
> I'm not proposing that, although I personally wouldn't mind it. I'm
> going to take over/fork a SQL-db editing tool, and am tempting to
> drop all support for quarantine in that.
> But before I did that, I wanted to see if there were good reasons to
> keep it I wasn't thinking of.
One of the big reasons for me is this: Why would you want to pass through
the 90%+ amount of email that is spam?
Why put this unnecessary load on internal systems and users?
We let the frontier antispam servers quarantine all blocked spam.
Users get a nightly report of all spam the day before. The report
first has an informational/instructional text, followed by a short list
of subject in all spam, and at the end a longer list consisting of
Timestamp, From, Subject and a clickable link for each blocked email.
The link will release the email from quarantine.
Sort of like what Maia Mailguard and others do, but developed in-house
and, IMHO, simpler to manage/use and less resource demanding.
We don't use any ham/spam feedback, we feel that users shouldn't have
to care about that (and most of them probably wouldn't either).
Objects in quarantine are autoerased when they are three months old.
We have done this for a couple of years, and we have no problems with
the user base (about 5000 with very varying degree of computer knowledge).
Peter Olsson pol@...