On Wed, Feb 25, 2015 at 02:46:29PM -0500, System Support wrote:
> > No, there should be no such delay because this is a single delivery
> > of a single message with two recipients.
> >
> > Note the identical smtp[7114] pids, with identical queue-id, delays
> > down to .01s and the same remote server queue id.
> >
> > You're not reading your logs correctly. Note also that your previous
> > 50/s is likely due to the default recipient limit, and we've just
> > been chasing ghosts.
>
> OK. Is there any way to rate limit the traffic to the relay for this use case?
> I tried reducing the default_destination_recipient_limit to 1 along with the
> relay and smtp versions.
You seem to not be paying attention. The deliveries *are* rate-
limited. You're reporting a *single* delivery. Naturally it is
not possible to rate limit just one delivery, rate limits apply
to multiple things that happen one after the other...
You need to send multiple messages to see evidence of rate limiting.
Email transaction (with 50 recipients):
EHLO <client>
MAIL FROM:<sender>
RCPT TO:<rcpt1>
RCPT TO:<rcpt2>
...
RCPT TO:<rcpt50>
DATA
From: <sender>
To: <some-list>
Subject: Announcing rate limits on things that happen once.
This single message has been rate limited, it will never
happen again
.
The above outlines a single message forwarded atomically to an SMTP
server. If the client were Postfix, it would log one log entry
for each of the 50 recipient addresses, recording that each was
delivered (with otherwise identical log information).
--
Viktor.
