I have been running my own mail server now for well over a decade. Whilst the actual physical hardware (or actually VPS system) may have changed once or twice during that time, the underlying software (postfix and dovecot on debian) has not really changed all that much. However, what has changed over the last decade or so, is the expectation that mail systems will be much more robust, better managed, less insecure (no more “open relays”) and harder on spam than had been the case in the early days of wide takeup of email by the public. Ignoring the “free” offerings from the likes of google, microsoft and others, it would arguably be cheaper, and certainly easier, for me to simply pay for an external mail service by one of the many providers out there. It is pretty easy to find companies offering to host personal email for about a tenner or at most twenty pounds a year. Those “solutions” (as providers seem to love to call their products) usually give you decent anti-spam, A/V scanning, POP3S/IMAPS connectivity (or if you really must, a webmail interface) and can usually alias mail to your preferred domain – particularly if you buy a domain name with your email service. But they always have limitations that I don’t like. The most obvious ones are: restrictions on the number of actual email addresses (as opposed to aliases), limited storage (though that is becoming less of a problem these days), and artificial restrictions on attachment sizes. And I’m bloody minded. I like to control my own email. I run my own email service for the same reason I manage my own DNS, run my own webservers, manage my own wordpress installation, run my own XMPP server and VPNs and manage my own domestic local network with assorted servers hanging off it. I like control and I dislike the opportunity outsourced services have for providing third parties access to my data. My personal data.
Besides, a boy needs a hobby.
However, I do occasionally get one or two problems in mail delivery – though usually /to/ my system rather than /from/ my system. For example I still get the occasional spam or cruddy email which gets past my protection mechanisms. Indeed I recently received one of those ridiculous extortion scam emails purporting to come from my own email adddress – more of which later – but this post is about an outbound mail failure from me to a friend of mine with a btinternet.com account.
I routinely correspond by email with a bunch of long standing friends who once lived relatively close together but are now more widely geographically dispersed. The group (or sub groups in some cases) get together on occasion for holidays, outings and meals. For some odd reason, many of those friends of mine have AOL accounts (I know, I know, but try telling them that). In a list of about two dozen regular correspondents, about a quarter of those people use AOL. The majority of the rest use BT, hotmail and gmail with one or two minor providers or work based accounts. On occasion in the past I have had mail to those AOL based accounts refused by AOL on the spurious grounds that my mail looked like spam because it was aimed at about half a dozen separate AOL accounts all at once. Well, that’s what happens when you “reply-all” to a mail list. Sadly AOL never could figure this out. After a while I gave up emailing their postmaster explaining the problem (and it was /their/ problem, identical email to the individual accounts always got through) because I never, ever, received a reply.
But this is about BT, not AOL.
Members of the mail list are shortly to meet for the group’s annual Christmas meal (it is always late, but hey) and one member “volunteered” to arrange the gathering, find a venue, sort menus etc. Said member has a btinternet email account (@btinternet.com) and he circulated a menu seeking choices for the meal. My reply was refused by BT with a “hard” 554 message which was reported to me by my mail system as below:
The mail system
email@example.com: host mx.bt.lon5.cpcloud.co.uk[184.108.40.206] said:
554 Message rejected on 2019/01/15 15:00:24 GMT, policy (220.127.116.11) – Your
message looks like SPAM or has been reported as SPAM please read
www.bt.com/bulksender (in reply to end of DATA command)
Now this was decidedly odd, because not 10 days beforehand I had happily sent earlier mails to the same address when our volunteer was initially talking about venue and proposed dates for the gathering. Just to be certain I wasn’t at fault, I checked the advice given by BT on their mail site referred to by the bounce message. Now the only thing I do not have set up for my mail server is DKIM signing. Everything else is hunky-dory – Proper “From” address? check. SPF? check. Proper MX records? check. Fixed IP address? check. PTR record? check. Good reputation? check. Not blacklisted? check (mxtoolbox says I’m fine). Furthermore, I never send HTML email (which I abhor as an abominable bastardisation of proper email standards) so did not have any embedded images or other bloody silly links in my mail). So after trying once or twice more later in the day (and failing) I emailed the BT postmaster saying I was having a problem and pointing out that whilst I might not use Domain keys, there seemed to me to be little else wrong with my email. I didn’t expect an answer, but you have to try,
BT responded – and they responded quickly. I sent my notification, with the failure message, to the BT postmaster address timed at 17.16. At 17.23 I received a reply saying:
Can you please send an example of the failing email to [investigation-address]@btinternet.com.
Please do not forward the email as an attachment but resend it.
Please let “postmaster” know when this has been sent so we can check the email’s content and possible reason for thinking it is spam.
Slightly stunned, I did as requested and a short time later (at half past midnight when I was asleep) I received another email from BT saying:
That email is scoring high as spam so I have reported it to our spam engine provider, I will email you again when I have some news.
Sure enough, that same morning at 02.50, I received the following good news:
We have made a change that should stop the emails being scored as spam, this change is being rolled-out now so please try again later.
On reading this when I got up that day I resent my email and, sure enough, it got through. Way to go BT! I have never, ever received that kind of rapid response from any ISP anywhere in the world – and I quite often email “abuse@” network addresses when some toerag or particularly persistent ‘bot shows up in my logs trying to do things I don’t like.
However, as much as I would like to believe that BT fixed a problem simply to accomodate my mail system, I actually think that unlikely. Given that mail from my system to @btinternet.com addresses had been working fine up until a few days ago, I think it much more likely that BT mail administrators had made some recent change, perhaps in one of their spam filters, which caused sigificant volumes of inbound mail to be rejected. My email had then simply been caught up in that wider problem and they were receiving queries or complaints from other mail administrators and not just me. Be that as it may, they still responded correctly, and efficiently as they moved to rectify whatever was causing the problem. So, my congratulations, and heartfelt thanks to the BT postmaster team for actually doing the sort of job that postmasters are supposed to, but rarely do properly.