Hello,
There was a discussion of DKIM and mail list back last October [1]. There was, and seems still is, an open mailman issue [2].
The Arch thread is informative, in particular around content-transfer-encoding. Perhaps this is helpful.
I believe I have read this thread already, but I will take a second look. As the mailman issue has yet to be resolved, there is not much which can be done here unfortunately.
Nope I don't mind -- 100% fine. I was just chucking at having the privilege to serve as the example :)
Well that is a relief. I am glad you got some enjoyment out of it :D
Just to say I may be out of date but I take ‘to spam’ to mean to send spam rather than classing ham as spam and spam-bucketing the email. :-)
People understood what I meant, no need to nitpick it :)
as you can see in the headers, the MTA sending mails is not mailman itself, but Postfix.
Mailman hooks into postfix, it still has its own logic. Hence why DKIM is being broken during the processing of the email.
As I got your mail via IPv6, I think it's just the default behaviour of Postfix:
http://www.postfix.org/postconf.5.html#smtp_address_preference
As pointed out in the docs, setting ipv6 as preference is considered unsafe. As an IPv6 enthusiast, I had set my Postfix to ipv6 myself and remember some issues with broken MX.
It seems most of my mail is delivered via IPv6, I have checked this but not with a large data set. But it seems most of the emails received from lists.archlinux.org is via IPv4, and the rare chance of IPv6 which is what I am finding is weird. https://serverfault.com/a/565123 This comment implies either Happy Eyeballs or a similar mechanism is being used by postfix. I have a friend who uses postfix and his mail server is configured dual stack and his mail ALWAYS uses IPv6, he hasn't set a preference either. Maybe Arch set IPv4 to be preferred and that is why?
But that note is rather confusing. The setting is describes as the address family to try *first*. Not as the only address family to try at all. So something in the docs is wrong. It's either the note or the documented behavior.
Maybe something to ask the postfix developers about, and possibly write a patch which better explains said mechanism? Thanks for the help! Take care, -- Polarian GPG signature: 0770E5312238C760 Website: https://polarian.dev JID/XMPP: polarian@icebound.dev