On 12 Feb 2023, at 18:35, Sabahattin Gucukoglu <listsebby@???> wrote:
> If I were to configure an SMTP transport in LMTP mode, can I cutthrough-deliver to it with an RCPT ACL? And what will happen if I have multiple recipients that have different post-data outcomes? I *think* it will work and should all go well, since only the first recipient matters, and that’s the one I care about, but it would be nice to clarify. It would be a neat little optimisation, to deliver directly into Cyrus and avoid the need to queue a delivery. If not, it’s not the end of the world since I don’t foresee any situation where the mail would be rejected anyway, but I’m curious.
So, for those who care, after hunting about, I find that this case is explicitly prohibited. See:
https://bugs.exim.org/show_bug.cgi?id=1576
(and the relevant bits of src/verify.c and src/acl.c)
Unfortunate, but at least it was anticipated. I certainly agree that PRDR makes it easier to proxy LMTP, or multi-recipient SMTP with PRDR on the other side. But it would have been quite nice to have a single recipient cut through, and only drop it with multi-recipient mail target-side, since in that case the workflow is basically similar to SMTP. Also, there is support for the reject extension in sieve that can only really be useful if you can proxy the rejection. Still, I don’t anticipate any savings on just telling lmtpd to simply swallow all validly-addressed mail. There’s no disk I/O saving either: proxying and spooling begin simultaneously. Is there a future for cutthrough or is it vestigial? Seems to be endangered by various features, such as CHUNKING.
Cheers,
Sabahattin
