On Tue, 3 Sep 2002, Alan J. Flavell wrote:
> On Tue, 3 Sep 2002, Hugh Sasse Staff Elec Eng wrote:
>
> > OK, so I can assume they are taking place but there have been no
> > failures yet. I expected something by now given the amount of
> > forged hotmail spam we get. Maybe that part of my expectations is
> > wrong :-)
>
> Hotmail definitely do reject bad user names... sometimes.
I tested it with an expect script on some old dud names, and
Hotmail did indeed reject them. [If people would like the script
I'm willing to share it
102 340 2434 /home/hgs/bin/smtpcallback
so it is not vast]
>
> I've seen all three of these behaviours from them in the past, in
> response to actual mail or bounces:
>
> - bad recipient reported at RCPT time
> - bad recipient reported at end of data
> - message accepted, even though known-bad recipient
>
> AIUI, only the first of these can be detected by callback, since it
?As I understand it?...
> proceeds no further than that into the SMTP transaction.
... agreed.
>
> Specimen extracted from exim4 log:
>
> sender verify fail for <bbg33617@???>: response to
> "RCPT TO:<bbg33617@???>" from mx09.hotmail.com [64.4.49.71] was
> 550 Requested action not taken:user account inactive
Ah, it didn't help that I was searching for "callback" :-)
an exigrep on everything for "response to" turned up nothing,
however, so it may just be that we have had none fail so far....
>
> You might want to note that they can also return 5xx to a callback
> when the sender's mailbox is full. (this may be a genuine user, so
> I'm ***-ing it out):
Thank you. That is worth knowing.
>
> sender verify fail for <***@hotmail.com>: response to
> "RCPT TO:<***@hotmail.com>" from mx01.hotmail.com [65.54.254.145] was
> 552 Requested action not taken: exceeded storage allocation
>
> We don't take any special action in that case, just treat that as
> grounds for rejection also.
I don't think I will. I have not got automated creation of local
blacklists yet, but as the problem of spam increases, so does the
temptation. Caveat(rope, self, hang)...
>
> But we already had this working in exim3, I can assure you: our
> configuration lines for v3 were e.g
>
> sender_verify_callback_domains = partial-dbm;/etc/exim/callback_domains.db
> sender_verify_hosts_callback = *
>
> which doesn't look significantly different from yours.
OK, thank you for this encouraging remark. I was sniffing around
for a log_smtp_callbacks option to increase the verbosity, to be
sure it was not succeeding silently :-), but I'm about to launch
into the brave new world of Exim4, so I don't think I'll worry about
this too much. Obviously given the existence of migration tools it
would be nice to be SURE it was working properly in my 3.36 config,
but I can live with this much uncertainty for a while.
>
> cheers
>
>
Thank you,
Hugh
