Perhaps so, but this is now what's happening out in the real world.
We also see the same behaviour from Yahoo (who started this before the
others a couple of years ago) and from Gmail, and as I said before,
our main client thought it was a good idea and did the same in their
SMTP-server. I think it would be a good idea to have Exim play nice
with these rules.
A correct implementation would be to say something with a message at
the FIRST email you decide to block, and then block the IP-address
after that. That way our SMTP-server would send as much as it could,
and then get blocked for a while - which would indicate to Exim that
the host is down (=retry rules in effect), but I understand why
they've done it like this - since they probably have some special
rules that would let a unique message slip by anyway, not to mention
the problem they might have to block an IP-address since they probably
have quite a large load-balancing setup with many entrypoints.
I can not change the way Microsoft, Yahoo, Gmail och others do
business - but I might tweak Exim to play better with their - somewhat
bad - rules.
Cheers!
/Charlie
2017-10-19 13:13 GMT+02:00 Evgeniy Berdnikov <bd4@???>:
> On Thu, Oct 19, 2017 at 11:19:53AM +0200, Charlie Elgholm wrote:
>> * 2017-10-18 11:42:23 1e4kjd-0006nY-4L SMTP error from remote mail
>> server after pipelined sending data block: 421 RP-001 (COL004-MC3F9)
>> Unfortunately, some messages from <ip> weren't sent. Please try again.
>> We have limits for how many messages can be sent per hour and per day.
>> You can also refer to
>> http://mail.live.com/mail/troubleshooting.aspx#errors.
>
> Microsoft guys are doing wrong thing. They should reject connections
> for specific ip-address, or return 4xx in place of greeting message
> and on MAIL, shuch errors could be treated as host-specific rejections.
> If status-codes 4xx are emitted on RCPT stage, such rejections are
> recipient-specific, so Exim could try other mails for the same MX.
>
> This log fragment looks like response to DATA, with pipelining on.
> It's a bad idea to put the ip-address rejection here.
> Legal 4xx response to DATA may be caused by problems with disk space,
> memory, system-wide filter (antivirus, ex) hangup on the receiver host.
> All these conditions are not related to specific recipient.
> So I think Exim havе better to treat 4xx on DATA as host-specific.
> --
> Eugene Berdnikov
--
Regards
Charlie Elgholm
Brightly AB
