Paul Tomblin via Postfix-users:
>
>
> On Sat, Jun 27, 2026, at 11:08 AM, Paul Tomblin via Postfix-users wrote:
> > On Sat, Jun 27, 2026, at 9:55 AM, Wietse Venema via Postfix-users wrote:
> >
> >> You may want to specify slow_smtp_destination_rate_delay to slow things
> >> further down.
> >
> > Ok, I made that one change (set it to 3s), did a postfix reload, and
> > now they've all got the process id, but the delay= changes every 5.
> > After the 2nd batch of 5, all the subsequent batches were deferred
> > (server temporarily unavailable). But it seems like when it retried,
> > the 3rd and 4th batches went through with delay=490 and delay=493
> > respectively, but everything else delayed again.
> >
>
> Oh, I spoke too soon. It appears after 4 retries, the final 19 of the members
> bounced. That's frustrating, because that looks like it's
> RoadRunner/Charter/Spectrum that's doing that.
>
> 2026-06-27T15:24:59.818522+00:00 mail postfix-slow/smtp[1587226]: D9D1182A6E:
> to=<[email protected]>,
> relay=pkvw-mx.msg.pkvw.co.charter.net[47.43.26.7]:25, delay=2293,
> delays=2290/2.8/0.33/0.13, dsn=5.1.0, status=bounced (host
> pkvw-mx.msg.pkvw.co.charter.net[47.43.26.7] said: 550 5.1.0 96.126.110.20
> blocked. Please see
> https://www.spectrum.net/support/internet/understanding-email-error-codes for
> more information. AUP#In-1300 (in reply to MAIL FROM command))
With slow_smtp_destination_rate_delay, the scheduler will enforce
a concurrency of 1 (deliver one message, pause, deliver one message,
etc.).
This means that you can safely set slow_smtp_destination_recipient_limit
to 1 (without rate delays, _destination_recipient_limit=1 will
actually increase concurrency. It's documented but not intuitive).
Fewer recipients spread out over lartger time intervals may be the
only way to deliver many recipients without trggering some rate
limit on their end. They might limit deliveries per hour, per day,
or per bigger interval, and they might change their mind.
Wietse
Wietse
_______________________________________________
Postfix-users mailing list -- [email protected]
To unsubscribe send an email to [email protected]
