> Publish forward-confirmed reverse DNS for every sending IP
> Every sending IP must have a reverse DNS entry pointing to a fully qualified domain name (FQDN). That hostname must, in turn, resolve back to the same sending IP address. This is known as forward-confirmed reverse DNS (FCrDNS). An IP with missing or mismatched reverse DNS is treated as suspicious and is one of the most common reasons for an IP to be considered for listing.
I self-host an email server behind my ISPs connection, and so I do not own the IP address that I send mail from, therefore I can't satisfy the reverse DNS requirement.
You and any other small installations won't be touched by this because you're not bulk senders. What's the duopoly here exactly?
You should be more concerned about your web hosts screwing you. GoDaddy et al routinely disable their customers' mailservers on spurious heuristic bases and with little recourse.
It takes quite a lot of ego to tell a stranger what they should be worried about. But, I’m curious… where do you get the idea that Spamhaus only cares about large senders? I’ve never seen that documented anywhere.
When I started as a SysAdmin, only 10 years ago, I was already taught that without a correct reverse DNS for your MX records, you will be blacklisted by most common blacklist or mail security (appliance) providers.
I ran a small mail server on AWS circa 2012 and it was best practice to do that even then and even on cloud. AWS had a process for correctly setting the reverse dns and everything.
No excuses really. If your IP owner wants you to send email they'll allow you to change their reverse dns entries. If they don't - you shouldn't be sending mail from them. Should have been enforced ages ago.
Yep. This has been best practice for well over a decade. Nothing in Spamhaus new requirements will stop e-mail from servers that have been properly set up.
Which is exactly where a lot of spam comes from, exploited systems that don't have correct reverse DNS. The checks are stringent because to allow exceptions is to allow people to abuse those exceptions.
FWIW you could probably get yourself a $5 VPS to use as the public portion of your mail setup that just forwards to your local mail server, which can in turn smart host through that VPS, and you'll have most the same features but with a dedicated IP that you can control both forward and reverse DNS through (depending on provider).
Depending on your ISP they may also allow for options for setting reverse DNS depending on your service. It's a bit niche these days unless you're paying for a business class service, but it really does depend on the ISP, as some offer it fairly cheap and some don't (independent ISPs may be more likely).
> Residential IPs seem to be the most trusted these days
Oh, I see we've got a stand-up comedian in our ranks. ;)
Most trusted ? You mean aside from all those residential IPs we see in our logs on a daily basis with the inevitably infected machines behind them taking part in DDoS attacks and other shenanigans.
At best residential IPs are in the same category of IPs of the so-called "bulletproof hosts" and similar.
Spamhaus and others maintain databases of residential IPs and whilst it doesn't get used for auto-blocking, most sensible organisations will reduce the trust score by a large chunk if its a residential IP.
My comment is based on the various recent discussions of "residential proxies", which are used for scraping because almost all of the VPS hosted IP ranges have got bad enough reputations that they're either blocked or rate limited or captcha-barricaded, etc.
I personally keep a log of IP addresses that I block, but I'm entirely blind to their source, and often get negative feedback about "but if it's residential then the IP address will be re-assigned to someone else" - which doesn't sound like a good "trust that traffic" argument to me.
I'm of the opinion that 99.9% of residential proxies are malware / involuntary, and that even the companies that attempt to 'herd' residential proxies legitimately are either criminal front companies or grew from criminal activity (not that that's always a bad thing).
Getting well off-topic here though.
P.S.
> Oh, I see we've got a stand-up comedian in our ranks. ;)
Spamhaus doesn't run anyone's mail server so that is what it is doing. If someone has configured their server to block mail over a certain Spamhaus score then that was their decision.
Yes, but Spamhaus is a trusted source of knowledge, and mailops tend to externalize some of the knowledge to Spamhaus, so their actions are not exactly "neutral".
Trusting Spamhaus at this point is a lot like trusting some random guy on Dread. You can't stop someone else from doing it, even though it's a bad idea.
I'm guessing good practice is renting a VPS and hoping it's not on blacklists already, or (more commonly) letting third parties read your mail by using a relay service. Or should we be applying for an ASN and buying IP blocks just to send mail?
There are SMTP relay services that are cheaper than a reliable VPS, that will manage monitoring the sending addresses for appearance in blocklists and do the work to fix that so you don't have to. I still send direct (my home account is in an almost-commercial plan with an ISP that takes things seriously including IP reputation), but if that starts to fail I'll move over to one of those services rather than using a cheap VPS for sending. Maybe more than one for resilience.
I used MXRoute for transactional messages from a project a while back and all worked well, and that service supports SMTP from your mail server as well as apps. I still have the account in case I need it again later. I've seen a few negative posts recently though, so look into that before diving in (a lot of it seems to be people skirting too close to the “no spamming” line, or obviously crossing it, and getting very unceremoniously kicked off, though IIRC it is essentially a one-man band and the accusation that the guy running it might have anger management issues could have a basis in truth). NameCrane is another mail delivery service that can operate as an outgoing smtp smarthost, that I considered for the above use, but I've never actually tried it (I intended to setup both, so one could act as a backup to remove a potential single-point-of-failure).
I don't (yet) have deliverability problems from my home IPv4 range, so I've not needed to use an external smarthost for general email yet and can't speak directly for quality of services when used in that manner, but I had no problems when using mxroute for transactional emails from a web-based app.
A reputable and conscientious VPS provider will rigorously block outbound port 25. Some of those allow you to request outbound port 25 on a discretionary basis. If they're quick to respond to complaints then their address blocks should be pretty clean. You can always lookup your assigned address in various blacklists and roll the dice again on a new instance if you get a bad one. Ditto if you find Google or Microsoft doesn't like it, either.
You should just host your mail server on a VPS and not overthink it. If someone is blocking your VPS and you urgently need to reach them, just use your old address on a big provider as a fallback.
My most productive day was shutting down the vps mail server and just paying 5 dollars a month. The undelivered outgoing and undelivered incoming mail was too random.
Undelivered incoming mail would be entirely your fault. Outgoing is where the spam filters can become a problem, but you should still try it and see for yourself because HN likes to exaggerate.
If you have a static IP address and they're allowing outbound port 25 then my first thought is that you might have some sort of premium residential or business account that permits hosting in the ToS, in which case I would then infer they might allow you to specify a PTR record in their zone or maybe even support delegation to you.
Lots of inferences there, but presuming Spamhaus isn't already blacklisting your IP address then that's additional evidence you don't have a vanilla residential IP address.
Have you asked, yet, about a PTR record? Not every ISP has a DNS system tied into a web portal for automated edits. Even some large ISPs still manange their zones through traditional zone files. But that doesn't mean they don't support modifying PTR records for customers. It may be just a matter of you putting in a request and some greybeard sysadmin resolving the ticket by firing up vi or emacs to edit one of their zone files. This might even be preferable because it's more likely they'd be open to delegating d.c.b.a.in-addr.arpa to your own DNS server. That would allow you to add and modify PTR records without having to go through them each time, whether manually or through some bespoke web service.
Haven't asked yet. I've got some near-future re-jigging of my email setup to do, so I'll be looking further into this.
DNS is something I've not self-hosted. It's currently delegated to Cloudflare, and has been for well over a decade (possibly pretty close to as long as Cloudflare has existed), with a handful of domains. For whatever reason, way back when, I just couldn't grok DNS zones etc., so threw my hands in the air. I think I'd probably be capable now, but have had a nostalgia-block due to the memory of the past lack of ability.
FWIW, you can host just the PTR record/zone and keep everything else as is. But probably just better to have the ISP serve the PTR record. Unless it's IPv6, then it might be useful to have a delegation for the subnet.
The majority of the internet has been blocking or silently discarding mail from your server then. This has been a de facto requirement for a long time.
If the ISP is not permitting you to configure the PTR of "your" addresses, then are they really your addresses?
Spamhaus specifically recommends a "branded" name. Deferring to the (recognizable as such) ISPs defaults, and thus raising the suspicion of attempting to wash all customers together is not terribly useful with other antispam systems either.
On residential, they're never "your" IP address. The one assigned to me is 'mostly static', but they did notify me that the IP address they'd assigned to my connection was going to change, which it now has.
Given my ISP has been kind enough to allow 'bridge mode' for my router, and outgoing traffic on port 25, then maybe I could ask them to change the reverse DNS lookup as well - feels like it may be a bridge too far, all I can do is ask though.
Separate, but related, I host email for multiple domains on this connection, so the reverse lookup would only be relevant for one of those domains. I'm not sure if there are additional ways around that, but haven't thought in too much detail about it yet.
> I host email for multiple domains on this connection, so the reverse lookup would only be relevant for one of those domains.
For that you'll need to pick a domain that you don't mind the others being associated with, and having all MX entries be pointing to the sending hostnames for that. Or you'll need multiple addresses and some way to route outgoing mail through the right one.
No, your ISP’s IPs are not “really your addresses”, and precisely no one claimed they were. This is a workaround so that someone can send emails out. Most people don’t have their own PI IP space, especially these days.
Yeah, it does have a reverse DNS that resolves to the ISP, and whilst looking at the article I had the same thought. Might see if I can implement that sooner rather than later.
> I self-host an email server behind my ISPs connection, and so I do not own the IP address that I send mail from, therefore I can't satisfy the reverse DNS requirement.
Me too, but you can pay my residential ISP a few euros per month extra for a static ipv4 allocation + custom PTR record. Maybe try sending them an email if they allow custom PTRs.
Yes, haven't had any problems, although I don't send a lot of email via my self hosted domains. Sent to work email, which is hosted by Microsoft, and to my personal Gmail accounts.
Interesting. For reasons did some nc to port 25 from a residential connection to a mid sized email provider and they just straight up refuse mail delivery from residential IPs, in the MAIL command stage even if I remember correctly.
In all reality though most spam that isn't trivially detected nowadays comes from Mailchimp, Sendgrid, etc., hacked email accounts (gmail, outlook, etc.).
I am unsure how the first ones are financially viable, maybe also hacked into?
Improperly setup mail servers I think are a thing of the past, since all the big providers have been blocking them for ages. Sure every now and then some come through, but at this point I believe these are basically all just old spam systems that are extremely easy to detect.
These techniques rely on it being hard (or at least long winded) to setup a domain and server properly, such that spammers (who are self selected to be lazy) won’t do it right and give up quickly.
But with AI, it’s easy to get DKIM, DMARC, HELO etc. set up perfectly for almost no effort (maybe fcrdns is a bit more complex as it’s a manual process at most ISPs)
I think we are in the end going to go back to it just being based on IP (v4) reputation, and this is all theatre.
This is not why these techniques are useful, though. They reduce plausible deniability and laundering-by-mixing. Because once you connected your unique (domain) name to your stuff, there is no "oh, it must have been that other customer that used that address before me" excuse any more. If everyone plays nicely and identifies their systems unambiguously, the question of who is responsible is clear. And thus, the reaction after detected misbehavior can be immediate and decisive.
DMARC, DKIM and SPF are not hard to deal with. HELO/PTR is considered basics.
If you need an AI to set these things up for you then you shold honestly consider having someone else run your email for you. Most hosters handle things like authentication out of the box.
Even before AI, was DKIM and DMARC hard to setup? It's been a streamlined process for at least four years now (as long as I've had to professionally enter records in a registrar), and even longer before that in my homelab setups.
I'm using Docker Compose on a Hetzner server that runs, among other things, Postfix, OpenDKIM and OpenARC.
A few weeks ago, I decided to use Claude Code to improve it. It found 2-3 misconfigured little things in Postfix and OpenDKIM. When it upgraded Postfix, I think it found the OpenARC main repo was deprecated. I mean, it worked, and none of them were security issues, but there's always something slightly wrong with email settings. And the only thing that Postfix does is forward emails to other mailboxes.
Yeah, most people would just geg any free bulk sender SaaS, and have everything set up perfectly in a few minutes, and have no technical background/just following step-by-step instructions.
In my experience, it's really only the homegrown email servers failing DKIM/DMARC in the current year, which must be simple ignorance. I've had mailcow for years and it handles DKIM/DMARC perfectly, and I assume Stalwart is exactly the same.
In the office, Google or Microsoft, it's been copy and pasting a couple records into your registrar for like four years. Not sure how admins are still messing this up even with AI now, haha.
None of these requirements are hard. Most of them were already requirements for any mail server with a spam filter. It's possible you have a very convoluted mail stack that makes this all very difficult to configure correctly, of course.
Gmail and Spamhaus bringing mail servers into the futuristic year of 2015 by requiring basic anti-spam measures is going to be better for everyone. Had they required scams like BIMI then I would've cared, but I'm almost disappointed that Spamhaus wasn't enforcing these requirements already to be honest.
Little known is that spamhaus isn't just for email. They also offer a list of good/bad domains, sometimes including personal blogs on domains that have never sent an email, making them ~inaccessible across entire countries, with ~no option to resolve the issue. (Ask me how I know.)
These requirements are standard practice. Half the internet, including all of the big providers, already rejects your mail if you don't meet them. Spamhaus is late to the party, probably because it doesn't actually have any way to check these since it isn't the one receiving the email.
Doesn't make them any better. They still exclude anyone who doesn't "own" their IP addresses. That does include a lot of compromised home PCs of course, but they could do better criteria.
> all of the big providers
All the big providers would of course prefer there were no small providers :)
> These requirements are standard practice.
Let's have some fun with that statement.
"Making war to have prisoners to sacrifice to the sun god is standard practice. Moctezuma, 1454."
I personally envision a future where I can just let a local LLM filter / categorize the E-Mail on-device where I can customize rules for me instead of having a certain racket do the "protection" for me.
I've been doing that for a while now. Works fantastically well, Qwen 3.8 got to very nearly 100% on my spam/ham benchmarks.
Please don't call Spamhaus a "racket". It's a service I've been relying on for many years now, and what they require now are basics that every service should fulfill.
My only problem is that we can't organize together and block Google/Amazon/Microsoft, where most of my spam comes from.
Anyone else having their email blocked by Microsoft and Spamhaus?
Microsoft Office 365 relies on Spamhaus and was blocking my email because the HELO on the mail server was mydomain.com instead of mailserver.mydomain.com.
Once the problem was fixed, it took time for Spamhaus to re-scan the mail server and take the mail server off the block list. And it took more time for Microsoft to stop blocking mail.
You can also reach out to sender support through the same, and they are ~reasonably responsive (though very clearly outsourced / limited power). If it's a lingering MS block they should be able to override it. I've had luck in the past (though this was on behalf of a notable tech company, I don't know that the support people cared at all about that fact).
SNDS is a complete joke. 2-3 times a year some or all of one of my subnets gets blocked by MS email services, outlook, hotmail, etc... SNDS never shows a single issue with any of verified subnets. It takes on average 1-2 weeks playing email tag with their support (who tell us to sign up for SNDS!) and them refusing to provide any evidence, before it magically just works again. It's happened so many times now I've begun to just not bother, and usually it still resolves itself. Meaning the support ticket is basically doing nothing. :-(
I've had a good experience with Spamhaus support who've put me on a whitelist despite my egress IP being listed in a UCLPROTECTL3. On the other hand getting anywhere close to de-listed on Microsofts domains, be it @hotmail.com, @outlook.com or anything else has been abysmal - Everything requires you to have some sort of Microsoft or EntraID login to even create a case, delist@microsoft.com is utterly unhelpful.
Yep, most are. Been setting up email servers for decades now. I've never once had that situation, only heard stories and I bet if that's the case then your provider will happily give you a new one.
Can someone clarify, do these rules apply only to people running their own mail transport?
Or does it also apply to people sending mail through their mail hosting company (when that company is not Google or Microsoft).
Specific example: I run servers "at home" that send mail through SMTP.
I have forward DNS properly set up, but my reverse DNS is syn-123-123-123-123.res.spectrum.com
My mail servers are, however, sending the mail through my mail host (DreamHost) and so I'm hoping that not having reverse DNS at home won't be a problem.
> requirement for every sending IP to have a valid HELO, a reverse DNS (PTR) record, and a matching A record.
How was this not the default? If you're going to send me mail, I already check all of those... and I also check to see if you hold a MX. If your DNS is _that_ bad, I probably don't want your spam anyways...
Every now and then I wonder how my mail inbox would feel when people would need to pay 1ct to send me an email, and how it would feel at 10ct and at 1usd.
It is not detecting spam anymore but imposing their rules on all mail servers that sucks a lot because I'm quite sure that the majority of mail servers with random HELO are legit senders. And no one cares about the HELO for good reasons for decade
The problem is that the sending mail server doesn't use Spamhaus but companies fighting for centralization and trying to get a monopoly on emails like Microsoft would use it and make you an outcast. That's the problem
Check the headers of messages received via Microsoft; they don't even care what your policy says and check the alignment and validity of your SPF/DKIM configuration just the same.
Ah. So DMARC can actually makes things less secure? Can't say I'm surprised. I've had spf and dkim signing set up since ~2017. I only set up dmarc in 2024.
When you say "through Google, Amazon, and Microsoft", is what you mean "through a server on GCP, AWS, or Azure" (or through their equivalent email-sending services)?
I assume you don't mean you're receiving a large volume of spam from @amazon.com or @microsoft.com email addresses?
I think we can assume that most people responding here are running their own mailserver that receives. I am. And I run a catch-all for my domain that doesn't check any spf/dkim/dmarc or any of that. And I can tell you, yes, the megacorps still send the most spam.
Are you suggesting that hosts along the path besides the sender and receiver mailserver are interpreting it and filtering? I'm not being glib, I just can't think of how what others do changes what my mailserver sees.
I mean, I get the whole "herd immunity" angle so that if everyone requires spf pass then spammers will start using spf. But this doesn't change anything about the ratio of megacorp spam versus independent spam.
Almost all my spam, similarly, is "legal spam". Any time you interact with any service you end up "agreeing" to be added to potentially hundreds of different mailing lists, and the "unsubscribe" link cheerfully tells you that you've been removed from Marketing Emails on Weekdays Odd Months Just For Accessories, but not the other ones. I have never in my life wanted to subscribe to anything.
> Publish forward-confirmed reverse DNS for every sending IP
> Every sending IP must have a reverse DNS entry pointing to a fully qualified domain name (FQDN). That hostname must, in turn, resolve back to the same sending IP address. This is known as forward-confirmed reverse DNS (FCrDNS). An IP with missing or mismatched reverse DNS is treated as suspicious and is one of the most common reasons for an IP to be considered for listing.
I self-host an email server behind my ISPs connection, and so I do not own the IP address that I send mail from, therefore I can't satisfy the reverse DNS requirement.
It's a stupid policy that will lead to many false positives among small installations, while having zero impact on actual spammers.
A lot of modern spam policing is just protectionism of a duopoly masking as security theatre.
Agree, Like airport security theater.
You and any other small installations won't be touched by this because you're not bulk senders. What's the duopoly here exactly?
You should be more concerned about your web hosts screwing you. GoDaddy et al routinely disable their customers' mailservers on spurious heuristic bases and with little recourse.
What does being (or not being) a bulk sender have to do with Spamhaus' requirements?
It takes quite a lot of ego to tell a stranger what they should be worried about. But, I’m curious… where do you get the idea that Spamhaus only cares about large senders? I’ve never seen that documented anywhere.
When I started as a SysAdmin, only 10 years ago, I was already taught that without a correct reverse DNS for your MX records, you will be blacklisted by most common blacklist or mail security (appliance) providers.
I ran a small mail server on AWS circa 2012 and it was best practice to do that even then and even on cloud. AWS had a process for correctly setting the reverse dns and everything.
No excuses really. If your IP owner wants you to send email they'll allow you to change their reverse dns entries. If they don't - you shouldn't be sending mail from them. Should have been enforced ages ago.
Yep. This has been best practice for well over a decade. Nothing in Spamhaus new requirements will stop e-mail from servers that have been properly set up.
Which is exactly where a lot of spam comes from, exploited systems that don't have correct reverse DNS. The checks are stringent because to allow exceptions is to allow people to abuse those exceptions.
FWIW you could probably get yourself a $5 VPS to use as the public portion of your mail setup that just forwards to your local mail server, which can in turn smart host through that VPS, and you'll have most the same features but with a dedicated IP that you can control both forward and reverse DNS through (depending on provider).
Depending on your ISP they may also allow for options for setting reverse DNS depending on your service. It's a bit niche these days unless you're paying for a business class service, but it really does depend on the ISP, as some offer it fairly cheap and some don't (independent ISPs may be more likely).
Don't those VPS IP ranges usually have bad reputation too?
I’m surprised your emails have ever gotten through if they’re coming from a residential IP. That’s definitely not considered good practice.
Residential IPs seem to be the most trusted these days, other than those, ironically, owned by the handful of big tech companies.
Email should be able to be run from anywhere.
> Residential IPs seem to be the most trusted these days
Oh, I see we've got a stand-up comedian in our ranks. ;)
Most trusted ? You mean aside from all those residential IPs we see in our logs on a daily basis with the inevitably infected machines behind them taking part in DDoS attacks and other shenanigans.
At best residential IPs are in the same category of IPs of the so-called "bulletproof hosts" and similar.
Spamhaus and others maintain databases of residential IPs and whilst it doesn't get used for auto-blocking, most sensible organisations will reduce the trust score by a large chunk if its a residential IP.
;)
My comment is based on the various recent discussions of "residential proxies", which are used for scraping because almost all of the VPS hosted IP ranges have got bad enough reputations that they're either blocked or rate limited or captcha-barricaded, etc.
I personally keep a log of IP addresses that I block, but I'm entirely blind to their source, and often get negative feedback about "but if it's residential then the IP address will be re-assigned to someone else" - which doesn't sound like a good "trust that traffic" argument to me.
I'm of the opinion that 99.9% of residential proxies are malware / involuntary, and that even the companies that attempt to 'herd' residential proxies legitimately are either criminal front companies or grew from criminal activity (not that that's always a bad thing).
Getting well off-topic here though.
P.S.
> Oh, I see we've got a stand-up comedian in our ranks. ;)
Says traceroute66. Love it.
But residential IP as a mail server has a very different trust score than residential IP as a scraper.
Why do you think they're ending up in your logs and not being blocked at an earlier layer?
> Why do you think they're ending up in your logs and not being blocked at an earlier layer?
Ah, but I didn’t say which logs :-)
I was also using ‘logs’ as a generic catch-all term.
> Email should be able to be run from anywhere.
And people should never send spam, but here we are.
Not being able to send mail at all from a residential IP is bullshit, though. Increase its spam score, but let it through.
Spamhaus doesn't run anyone's mail server so that is what it is doing. If someone has configured their server to block mail over a certain Spamhaus score then that was their decision.
Yes, but Spamhaus is a trusted source of knowledge, and mailops tend to externalize some of the knowledge to Spamhaus, so their actions are not exactly "neutral".
Trusting Spamhaus at this point is a lot like trusting some random guy on Dread. You can't stop someone else from doing it, even though it's a bad idea.
I'm guessing good practice is renting a VPS and hoping it's not on blacklists already, or (more commonly) letting third parties read your mail by using a relay service. Or should we be applying for an ASN and buying IP blocks just to send mail?
There are SMTP relay services that are cheaper than a reliable VPS, that will manage monitoring the sending addresses for appearance in blocklists and do the work to fix that so you don't have to. I still send direct (my home account is in an almost-commercial plan with an ISP that takes things seriously including IP reputation), but if that starts to fail I'll move over to one of those services rather than using a cheap VPS for sending. Maybe more than one for resilience.
What's a good SMTP service? Sendgrid is too $s
I used MXRoute for transactional messages from a project a while back and all worked well, and that service supports SMTP from your mail server as well as apps. I still have the account in case I need it again later. I've seen a few negative posts recently though, so look into that before diving in (a lot of it seems to be people skirting too close to the “no spamming” line, or obviously crossing it, and getting very unceremoniously kicked off, though IIRC it is essentially a one-man band and the accusation that the guy running it might have anger management issues could have a basis in truth). NameCrane is another mail delivery service that can operate as an outgoing smtp smarthost, that I considered for the above use, but I've never actually tried it (I intended to setup both, so one could act as a backup to remove a potential single-point-of-failure).
I don't (yet) have deliverability problems from my home IPv4 range, so I've not needed to use an external smarthost for general email yet and can't speak directly for quality of services when used in that manner, but I had no problems when using mxroute for transactional emails from a web-based app.
AWS SES, set it all up on your account with wraps cli or with your agent. It is the cheapest option.
I've used and recommended DuoCircle for well over a decade.
Take a look at AnyMXRelay from AnyDomain
A reputable and conscientious VPS provider will rigorously block outbound port 25. Some of those allow you to request outbound port 25 on a discretionary basis. If they're quick to respond to complaints then their address blocks should be pretty clean. You can always lookup your assigned address in various blacklists and roll the dice again on a new instance if you get a bad one. Ditto if you find Google or Microsoft doesn't like it, either.
You can ask most reputable VPS providers to open port 25 explaining what domain you'll be using, expected volume, etc...
You should just host your mail server on a VPS and not overthink it. If someone is blocking your VPS and you urgently need to reach them, just use your old address on a big provider as a fallback.
My most productive day was shutting down the vps mail server and just paying 5 dollars a month. The undelivered outgoing and undelivered incoming mail was too random.
Undelivered incoming mail would be entirely your fault. Outgoing is where the spam filters can become a problem, but you should still try it and see for yourself because HN likes to exaggerate.
Who did you go with?
If you have a static IP address and they're allowing outbound port 25 then my first thought is that you might have some sort of premium residential or business account that permits hosting in the ToS, in which case I would then infer they might allow you to specify a PTR record in their zone or maybe even support delegation to you.
Lots of inferences there, but presuming Spamhaus isn't already blacklisting your IP address then that's additional evidence you don't have a vanilla residential IP address.
My ISP does block outbound port 25 by default, I specifically asked for it to be unblocked for me, and they obliged.
Have you asked, yet, about a PTR record? Not every ISP has a DNS system tied into a web portal for automated edits. Even some large ISPs still manange their zones through traditional zone files. But that doesn't mean they don't support modifying PTR records for customers. It may be just a matter of you putting in a request and some greybeard sysadmin resolving the ticket by firing up vi or emacs to edit one of their zone files. This might even be preferable because it's more likely they'd be open to delegating d.c.b.a.in-addr.arpa to your own DNS server. That would allow you to add and modify PTR records without having to go through them each time, whether manually or through some bespoke web service.
Haven't asked yet. I've got some near-future re-jigging of my email setup to do, so I'll be looking further into this.
DNS is something I've not self-hosted. It's currently delegated to Cloudflare, and has been for well over a decade (possibly pretty close to as long as Cloudflare has existed), with a handful of domains. For whatever reason, way back when, I just couldn't grok DNS zones etc., so threw my hands in the air. I think I'd probably be capable now, but have had a nostalgia-block due to the memory of the past lack of ability.
FWIW, you can host just the PTR record/zone and keep everything else as is. But probably just better to have the ISP serve the PTR record. Unless it's IPv6, then it might be useful to have a delegation for the subnet.
No you can't host the PTR record yourself unless either you own the IP address block or you get your ISP to CNAME the record to you.
For 15+- years it’s common practice to refuse any mails from a Server where the glue records don’t match.
The majority of the internet has been blocking or silently discarding mail from your server then. This has been a de facto requirement for a long time.
Doesn't make it any better.
My residential connection has working reverse DNS, although I fail to see how this has anything to do with spam.
Can you purchase forward IP from non-ISP provider separately?
Is IPv6 roaming a thing yet (after 25 years)?
I think you can have a simple wiregaurd setup in small linode or so and then have ptr records there
Most ISPs have a PTR that also resolves to the IP; try using the one they chose for you (cust-462728.atl1.ispname.com) in your HELO.
But Spamhaus also wants the HELO domain to have a valid SPF record, which ISP PTR domains cannot satisfy.
If the ISP is not permitting you to configure the PTR of "your" addresses, then are they really your addresses? Spamhaus specifically recommends a "branded" name. Deferring to the (recognizable as such) ISPs defaults, and thus raising the suspicion of attempting to wash all customers together is not terribly useful with other antispam systems either.
On residential, they're never "your" IP address. The one assigned to me is 'mostly static', but they did notify me that the IP address they'd assigned to my connection was going to change, which it now has.
Given my ISP has been kind enough to allow 'bridge mode' for my router, and outgoing traffic on port 25, then maybe I could ask them to change the reverse DNS lookup as well - feels like it may be a bridge too far, all I can do is ask though.
Separate, but related, I host email for multiple domains on this connection, so the reverse lookup would only be relevant for one of those domains. I'm not sure if there are additional ways around that, but haven't thought in too much detail about it yet.
> I host email for multiple domains on this connection, so the reverse lookup would only be relevant for one of those domains.
For that you'll need to pick a domain that you don't mind the others being associated with, and having all MX entries be pointing to the sending hostnames for that. Or you'll need multiple addresses and some way to route outgoing mail through the right one.
Cheers, will look into that.
No, your ISP’s IPs are not “really your addresses”, and precisely no one claimed they were. This is a workaround so that someone can send emails out. Most people don’t have their own PI IP space, especially these days.
Yeah, it does have a reverse DNS that resolves to the ISP, and whilst looking at the article I had the same thought. Might see if I can implement that sooner rather than later.
Worth calling and asking, some ISPs will update your static IPs rDNS on request.
> I self-host an email server behind my ISPs connection, and so I do not own the IP address that I send mail from, therefore I can't satisfy the reverse DNS requirement.
Me too, but you can pay my residential ISP a few euros per month extra for a static ipv4 allocation + custom PTR record. Maybe try sending them an email if they allow custom PTRs.
Tell your ISP to fix their DNS then. It doesn't have to be your.domain, it can be any.domain, as long as RDNS and DNS match.
Are you able to actually get emails delivered?
Yes, haven't had any problems, although I don't send a lot of email via my self hosted domains. Sent to work email, which is hosted by Microsoft, and to my personal Gmail accounts.
Interesting. For reasons did some nc to port 25 from a residential connection to a mid sized email provider and they just straight up refuse mail delivery from residential IPs, in the MAIL command stage even if I remember correctly.
Yeah, well, you can’t always get what you want.
In all reality though most spam that isn't trivially detected nowadays comes from Mailchimp, Sendgrid, etc., hacked email accounts (gmail, outlook, etc.).
I am unsure how the first ones are financially viable, maybe also hacked into?
Improperly setup mail servers I think are a thing of the past, since all the big providers have been blocking them for ages. Sure every now and then some come through, but at this point I believe these are basically all just old spam systems that are extremely easy to detect.
These techniques rely on it being hard (or at least long winded) to setup a domain and server properly, such that spammers (who are self selected to be lazy) won’t do it right and give up quickly.
But with AI, it’s easy to get DKIM, DMARC, HELO etc. set up perfectly for almost no effort (maybe fcrdns is a bit more complex as it’s a manual process at most ISPs)
I think we are in the end going to go back to it just being based on IP (v4) reputation, and this is all theatre.
This is not why these techniques are useful, though. They reduce plausible deniability and laundering-by-mixing. Because once you connected your unique (domain) name to your stuff, there is no "oh, it must have been that other customer that used that address before me" excuse any more. If everyone plays nicely and identifies their systems unambiguously, the question of who is responsible is clear. And thus, the reaction after detected misbehavior can be immediate and decisive.
DMARC, DKIM and SPF are not hard to deal with. HELO/PTR is considered basics.
If you need an AI to set these things up for you then you shold honestly consider having someone else run your email for you. Most hosters handle things like authentication out of the box.
Even before AI, was DKIM and DMARC hard to setup? It's been a streamlined process for at least four years now (as long as I've had to professionally enter records in a registrar), and even longer before that in my homelab setups.
I'm using Docker Compose on a Hetzner server that runs, among other things, Postfix, OpenDKIM and OpenARC.
A few weeks ago, I decided to use Claude Code to improve it. It found 2-3 misconfigured little things in Postfix and OpenDKIM. When it upgraded Postfix, I think it found the OpenARC main repo was deprecated. I mean, it worked, and none of them were security issues, but there's always something slightly wrong with email settings. And the only thing that Postfix does is forward emails to other mailboxes.
Yeah, most people would just geg any free bulk sender SaaS, and have everything set up perfectly in a few minutes, and have no technical background/just following step-by-step instructions.
Hell even BIMI isn't very hard these days.
It is pretty straightforward. If you can learn regex or cron expressions, you can learn their meaning and format.
It's gotten even easier with newer tooling. Stalwart will manage most of it for you if you want for example.
No AI required.
In my experience, it's really only the homegrown email servers failing DKIM/DMARC in the current year, which must be simple ignorance. I've had mailcow for years and it handles DKIM/DMARC perfectly, and I assume Stalwart is exactly the same.
In the office, Google or Microsoft, it's been copy and pasting a couple records into your registrar for like four years. Not sure how admins are still messing this up even with AI now, haha.
HELO hostname / banner is a little tricker
I expect someone attempting to make money by sending mass emails does not have the ability to create DNS records or figure it out alone
None of these requirements are hard. Most of them were already requirements for any mail server with a spam filter. It's possible you have a very convoluted mail stack that makes this all very difficult to configure correctly, of course.
Gmail and Spamhaus bringing mail servers into the futuristic year of 2015 by requiring basic anti-spam measures is going to be better for everyone. Had they required scams like BIMI then I would've cared, but I'm almost disappointed that Spamhaus wasn't enforcing these requirements already to be honest.
Spamhaus is effectively a protection racket. However, these specific requirements are extremely reasonable.
Little known is that spamhaus isn't just for email. They also offer a list of good/bad domains, sometimes including personal blogs on domains that have never sent an email, making them ~inaccessible across entire countries, with ~no option to resolve the issue. (Ask me how I know.)
Suing is usually an option. The problem is that it costs a few hundred dollars minimum, to even find out whether it's an option.
No they aren't. As it's been said elsewhere, they basically exclude all home email servers.
The road to hell is paved with good intentions. I mean if those major blacklist organizations have a shred of good intentions left.
These requirements are standard practice. Half the internet, including all of the big providers, already rejects your mail if you don't meet them. Spamhaus is late to the party, probably because it doesn't actually have any way to check these since it isn't the one receiving the email.
Doesn't make them any better. They still exclude anyone who doesn't "own" their IP addresses. That does include a lot of compromised home PCs of course, but they could do better criteria.
> all of the big providers
All the big providers would of course prefer there were no small providers :)
> These requirements are standard practice.
Let's have some fun with that statement.
"Making war to have prisoners to sacrifice to the sun god is standard practice. Moctezuma, 1454."
I guess Godwin's law needs an update :)
We already tried your way, it didn't work.
I personally envision a future where I can just let a local LLM filter / categorize the E-Mail on-device where I can customize rules for me instead of having a certain racket do the "protection" for me.
I've been doing that for a while now. Works fantastically well, Qwen 3.8 got to very nearly 100% on my spam/ham benchmarks.
Please don't call Spamhaus a "racket". It's a service I've been relying on for many years now, and what they require now are basics that every service should fulfill.
My only problem is that we can't organize together and block Google/Amazon/Microsoft, where most of my spam comes from.
I personally just envision a world where I only permit emails from sends that I personally authorise.
Whitelist government utilities and family, but everything else needs to be authorised using some yet to be drafted RFC.
Life is complicated already. 3 legitimate emails a week max seems sensible.
Prompt injection. Just like Bayesian poisoning back in the day.
Anyone else having their email blocked by Microsoft and Spamhaus?
Microsoft Office 365 relies on Spamhaus and was blocking my email because the HELO on the mail server was mydomain.com instead of mailserver.mydomain.com.
Once the problem was fixed, it took time for Spamhaus to re-scan the mail server and take the mail server off the block list. And it took more time for Microsoft to stop blocking mail.
https://check.spamhaus.org/
Make sure to register on their SNDS and you might be able to find the issue? https://substrate.office.com/ip-domain-management-snds/snds
You can also reach out to sender support through the same, and they are ~reasonably responsive (though very clearly outsourced / limited power). If it's a lingering MS block they should be able to override it. I've had luck in the past (though this was on behalf of a notable tech company, I don't know that the support people cared at all about that fact).
SNDS is a complete joke. 2-3 times a year some or all of one of my subnets gets blocked by MS email services, outlook, hotmail, etc... SNDS never shows a single issue with any of verified subnets. It takes on average 1-2 weeks playing email tag with their support (who tell us to sign up for SNDS!) and them refusing to provide any evidence, before it magically just works again. It's happened so many times now I've begun to just not bother, and usually it still resolves itself. Meaning the support ticket is basically doing nothing. :-(
I've had a good experience with Spamhaus support who've put me on a whitelist despite my egress IP being listed in a UCLPROTECTL3. On the other hand getting anywhere close to de-listed on Microsofts domains, be it @hotmail.com, @outlook.com or anything else has been abysmal - Everything requires you to have some sort of Microsoft or EntraID login to even create a case, delist@microsoft.com is utterly unhelpful.
Surely working/aligned forward/reverse DNS has been mostly required for SMTP for years?
It's pretty much the only thing I usually bother with reverse DNS for.
Is it possible anymore to even acquire an IP address that’s clean and use it to self host?
Yep, most are. Been setting up email servers for decades now. I've never once had that situation, only heard stories and I bet if that's the case then your provider will happily give you a new one.
Can someone clarify, do these rules apply only to people running their own mail transport?
Or does it also apply to people sending mail through their mail hosting company (when that company is not Google or Microsoft).
Specific example: I run servers "at home" that send mail through SMTP.
I have forward DNS properly set up, but my reverse DNS is syn-123-123-123-123.res.spectrum.com
My mail servers are, however, sending the mail through my mail host (DreamHost) and so I'm hoping that not having reverse DNS at home won't be a problem.
Am I reading the Spaumhaus rules correctly?
Since it's for the outbound mail server how would it possibly apply to any other step in the chain?
This may be a new requirement by spamhaus, but this is an old school standard config practice for serving email...
> requirement for every sending IP to have a valid HELO, a reverse DNS (PTR) record, and a matching A record.
How was this not the default? If you're going to send me mail, I already check all of those... and I also check to see if you hold a MX. If your DNS is _that_ bad, I probably don't want your spam anyways...
so can I get COI enforced and bad actors actually punished, or is this just gunna smash the little guys again?
can i get recruiter list spam punished?
Every now and then I wonder how my mail inbox would feel when people would need to pay 1ct to send me an email, and how it would feel at 10ct and at 1usd.
Currently it seems to cost 0.001ct.
There are actual solutions doing something close to that. Hashcash, the grandpa of PoW systems for example is really easy to implement.
http://hashcash.org/
However, advertisers and probably organizations like Spamhouse would be very sad if that was more widely used.
It is not detecting spam anymore but imposing their rules on all mail servers that sucks a lot because I'm quite sure that the majority of mail servers with random HELO are legit senders. And no one cares about the HELO for good reasons for decade
Then don't use Spamhaus. Spamhaus hasn't been serious for a long while.
The problem is that the sending mail server doesn't use Spamhaus but companies fighting for centralization and trying to get a monopoly on emails like Microsoft would use it and make you an outcast. That's the problem
So don't send emails to people on Microsoft. Or do, but when they don't receive it, blame them for using a service that can't even receive an email.
I checked my site with Codex and it failed due to Cloudflare and Resend.
Being picky about what is sent in HELO/EHLO seems like useless nitpicking toil that will not prevent any spam.
Not by itself. But it solves the "that was not me" dance: It was you. Your zone points there = your responsibility!
>Do not remain at p=none permanently. Starting with p=none is acceptable, but we recommend having an enforcement policy.
I've published a DMARC record that basically says to ignore DMARC and do nothing. I wonder if this will be enough.
I have no plans to ever change from none.Given that it's enough for providers like Gmail, etc. I strongly assume so.
If you have SPF/DKIM active this setting will void them.
Check the headers of messages received via Microsoft; they don't even care what your policy says and check the alignment and validity of your SPF/DKIM configuration just the same.
This is just not correct.
Ah. So DMARC can actually makes things less secure? Can't say I'm surprised. I've had spf and dkim signing set up since ~2017. I only set up dmarc in 2024.
Most of my spam is through Google, Amazon and Microsoft. They don't care: they are "too big to fail" at this point. Nobody will block Google, right?
We've allowed them to grow this way by preferring convenience.
When you say "through Google, Amazon, and Microsoft", is what you mean "through a server on GCP, AWS, or Azure" (or through their equivalent email-sending services)?
I assume you don't mean you're receiving a large volume of spam from @amazon.com or @microsoft.com email addresses?
I get plenty of spam from @gmail.com addresses, yes
@gmail.com and @outlook.com (or whatever the latest Microsoft free email domains are) are pretty common to get spam from.
It's from @gmail.com
> Most of my spam is through Google, Amazon and Microsoft
Survivorship bias
I think we can assume that most people responding here are running their own mailserver that receives. I am. And I run a catch-all for my domain that doesn't check any spf/dkim/dmarc or any of that. And I can tell you, yes, the megacorps still send the most spam.
Nothing you said negates my point
You do not exist in a vacuum.
Are you suggesting that hosts along the path besides the sender and receiver mailserver are interpreting it and filtering? I'm not being glib, I just can't think of how what others do changes what my mailserver sees.
I mean, I get the whole "herd immunity" angle so that if everyone requires spf pass then spammers will start using spf. But this doesn't change anything about the ratio of megacorp spam versus independent spam.
Can you be more explicit?
What?
Almost all my spam, similarly, is "legal spam". Any time you interact with any service you end up "agreeing" to be added to potentially hundreds of different mailing lists, and the "unsubscribe" link cheerfully tells you that you've been removed from Marketing Emails on Weekdays Odd Months Just For Accessories, but not the other ones. I have never in my life wanted to subscribe to anything.
wait, these are all only just now new? what has taken them so long?