Skip to content

IPv6-only hosting: who actually runs production here?

VPS Hosting by mediaaustin 14 replies 3.6K views
#11

And I have been fighting it for a month now. Some big providers still greylist based on missing A records which is absurd but here we are. Also

olya said:
The provided /64 is routed correctly but documentation is wrong about subnet sizing
— which provider is this? I have seen the same issue on GreenCloudVPS where their panel says /56 but routing table shows /64. Took a ticket to clarify. Thread has been quiet for a while, curious if people stuck with it or went back to dual-stack.

#12

147 days and nobody called out Ricardo77 for that word salad

mediaaustin said:
The trick is setting resolver to a v6-only DNS so you do not accidentally leak A queries

What resolver are you actually using? I've been tempted to try this on Contabo but their v6 DNS still returns A records for some internal health checks and I do not trust it

Also did you ever get Postfix working acceptably or did you punt to a relay

#13

214 days, guess nobody else took the plunge.

mediaaustin said:
Email deliverability is the real battlefield

Did you ever solve outbound? I tried relaying through GreenCloudVPS's v6 smarthost but major receivers still greylist anything without a v4 PTR. Ended up tunneling to a legacy IP for mail specifically, which feels like cheating.

Also their console finally fixed the subnet docs, sort of. Still need that bastion though.

licensing is a suggestion
#14

238 days, nice. I have inbound sorted but every time I try to cut the v4 cord on submission I get bitten by some provider's broken 6to4 or a middlebox that eats EHLO. Curious if mediaaustin ever finished that thought or if it is still the same battlefield.

#15

Did anyone actually get Postfix working reliably on v6-only?

mediaaustin said:
Email deliverability is the real battlefield
And I have been fighting it for months.

I moved a personal mailserver to GreenCloudVPS v6-only last year. Outbound to Gmail and Outlook works maybe 60 percent of the time. The rest soft-bounce with vague "network error" messages that I am pretty sure mean their edge does not like something about my HELO or reverse DNS.

I have PTR set through their panel, SPF and DKIM passing, but I keep wondering if the problem is simply that some receiving MXs still do not publish AAAA. So my resolver tries, gets nothing, and Postfix gives up instead of falling back. I tried forcing inet_protocols = ipv6 but that made it worse.

Curious if you found a specific Postfix main.cf incantation that helps, or if you just accept some destinations are unreachable from v6-only.

Post a reply

You need an account to reply. Log in or register to join the conversation.

Post reply Preview Save draft