On Wed, Sep 25, 2019 at 1:23 PM Bill Cole via mailop <mailop@mailop.org> wrote:
> On 25 Sep 2019, at 5:18, Simon Lyall via mailop wrote: > > > Just had a bunch of people at a domain get unsubscribed from this > > list. > > Something is very wrong there... > > 4xx failures, even persistent ones that ultimately fail the message for > queue timeout, should not cause an instant unsub from one message > failure. Are your bounce handling settings extra-strict? > Mailman's bounce handling behavior has always been kind of overly strict, that's why it had so much trouble with DMARC rejections, for instance. And you do want your bounce handling to remove users who "always" fail with a 4xx, though when/how you do that is probably more art than science. Ditto for various 5xx errors, ie content based failures may be specific to a single message, but may also apply to every message from the list. > Appears to be some weird Google rule (which probably made sense with > > they were not the MX for 30% of all active domains) > > > > Any chance of them fixing it > > It's not broken. It's functionally identical to sites that enforce a > one-RCPT-per-DATA rule. > > The right behavior when a MX gives a 4xx response to RCPT is for the > sending MTA to continue with any remaining recipients, remove the > successful ones from the envelope, and requeue the message with the > 4xx-ed recipients later. Repeat as necessary to clear the envelope or > until you hit a maximum queue lifetime. > At one point, we only used multiple recipients on the first attempt, splitting up all of the retries into individual tasks as a work around for various things like this. YMMV. Brandon
_______________________________________________ mailop mailing list mailop@mailop.org https://chilli.nosignal.org/cgi-bin/mailman/listinfo/mailop
