blank "is" report lines in syslog?
Matt Kettler
mkettler at evi-inc.com
Fri Jan 27 23:51:21 GMT 2006
DAve wrote:
>> You're in the same basic place that many mail operators were in when
>> spammers
>> discovered making use of open relays. At the time open relays were
>> commonly
>> accepted.
>
>
> Why would that be? I'm not an open relay. If you connect to one of my
> MailScanner boxes, milter-ahead will check for the existance of the
> recipient account on my toasters, and accept or reject the connection.
> How does that make me an open relay?
Well then you don't have the problem I'm talking about.
When you said you "weren't blind queuing" it was unclear you understood what I
meant. Some people claim that checking the domain only is enough to be "non blind".
> A message comes in on MailScanner box #1, it is a 15mb attachment. It takes awhile to get through MailScanner and be delivered to the toaster, and for the toaster to update the users quota. But Milter-ahead checked before the message was accepted, and there is room to deliver the message. Unfortunately, MailScanner box #2 has a 3mb message that gets through a bit faster, lands on the toaster, and gets delivered. Now when MailScanner box #1 tries to deliver, the message won't fit within the quota. KAPOW!, the message is bounced. This has happened as well, to me.
Are you sure milter-ahead verifies the storage at all? I did not realize it had
this ability. However, I originally said that insufficient queue was a
reasonable case to cause post-delivery bounces.
> post-smtp bounces can and will happen.
Agreed.. They only shouldn't happen for nonexistent users, except the
corner-case of arrives-at-instant of deletion.
More information about the MailScanner
mailing list