Martin Husemann wrote in <20241009091049.gd12...@mail.duskware.de>: ... |Greylisting is harmless and only causes a very small delay in mail \ |delivery. |It needs no manual intervention.
Except that some do not get it right. And to mention that NetBSD was known for notorious *long* (as in, many hours, for a normal postfix queue, many tries) gray list periods. I always mention netbsd.org when somebody has problems, just recently for example a RedHat employee, it turned out they get email services via mimecast.com, and they talk via bug tracker, and then you get R 190 Customer Reply 2024-10-02 20:32 2697/ 167169 ┕▸ with top posting and HTML, of course, and they say The Greylisting standards that Mimecast applies are RFC compliant. The following is how Mimecast applies greylisting on both inbound and outbound messaging: [Image_2024-10-02_14-32-12.png] You can have this 120 KB picture if you want! It practically gives some words regarding RFC 6647, with a note that normally Exchange servers (yay!!) retries posting every 10 minutes. P.S.: with mimecast.com the problem on their side seems to be something different, i said (i did not open the customer issue, i was cc:d at one time in this lengthy thing) |Nice. But since Tomáš *has passed* grey listing as (he is in my |DB, as shown in the thread), it can only be sender address |verification -- "callback verification" [1] as wikipedia calls it. | | [1] https://en.wikipedia.org/wiki/Callback_verification Silence ever since (likely i was taken out, i was willing to un-white/allowlist them again so we could test that, but, then not). But many, many more have problems with graylisting, and do not retry etc etc, so you better watch out when you get in first contact. Ie notorious email-service-outsourced users like German government (local city stuff and such), hospitals, whatever. That is *i* will not agree that "harmless" of yours. I for myself should not have written my graylisting thing, but instead try to upstream to postfix an extension of his reject_unverified_sender, ie the mentioned, it would have taken a bit only -- ie, i use delay-max 300 delay-min 0 count 1 focus-sender (the latter being my invention), and that gets it in practice, but if upstreamed we would reject once without the bit, and only truly verify thereafter. Ok two bits, as you want to remember whether verification succeeded. But so it is. --steffen | |Der Kragenbaer, The moon bear, |der holt sich munter he cheerfully and one by one |einen nach dem anderen runter wa.ks himself off |(By Robert Gernhardt)
