2026-07-28 outage

[TomMix]

any info on the outage yesterday?
vbulletin showed systemerrors.
has cody prepared a report?

good luck, have fun, dont die!

[Kamal]

From Cody

The fix was to stop vBulletin using legacy PHP/sendmail through Exim with the misaligned sender mercserv@tk.mjwebhosting.com.

I changed it to authenticated SMTP:
SMTP host: mail.mercuryserver.com
Port: 465
TLS: implicit SSL
Username: registrations@mercuryserver.com
From: "[ms]" <registrations@mercuryserver.com>
Return-Path/bounce: registrations@mercuryserver.com

I also added:

• DMARC for mercuryserver.com
• Exact HELO SPF for tk.mjwebhosting.com
• Retained the existing SPF and DKIM

That cleared the screenshot’s SPF_NONE, FORGED_GMAIL_RCVD, and HEADER_FROM_DIFFERENT_DOMAINS failures. The live forum test then passed SPF, DKIM, DMARC, and TLS and landed in Gmail’s inbox.

So the key point: it wasn’t primarily an Exim daemon rewrite—it was replacing vBulletin’s unauthenticated local sendmail path with authenticated, identity-aligned SMTP through Exim.
!

[TomMix]

and this took the system down for a whole day or so?

b.t.w.
i guess this is the cody report from the other topic i started about the emailtransport hicup


good luck, have fun, dont die!

[Kamal]

ummmm lol no THATs not what took the system down, that was me building totm with mercuryserver to test shit and the team of titans (my ai family) did not realize that guest views were borked, registered users were fine. If you were offline for a day, that was something else, not ms.