Hacker Newsnew | past | comments | ask | show | jobs | submit | horusthegame's commentslogin

> You can perfectly self-host while using an outgoing SMTP provider.

That's mostly selfhosting, not fully selfhosting. I full selfhost my email, I 100% control as much of it as is possible to do. When I send an email to someone, it goes from my client to my server and out to the recipient's server.


Still, I don't understand why someone who was self-hosting for 23 years suddenly abandons everything, instead of going "mostly self-hosting".

I genuinely think some people are not aware you can have free, reliable, outgoing SMTP providers.


How often do you check TLS certs the recipient' server provides?

And you can have an additional route just for some recipients, that doesn't make it less (or more) selfhosting.


> The odds that this works is basically zero.

It's times like this that I wish HN had better moderation, you are insultingly rude to the parent and to one of my comments.


While you may not have liked the way it was said, this users reply is actually responding with correct information.

While setting up your own mail server is fun, rewarding, and pretty easy, there are serious problems that anyone considers it needs to be made aware of. Over any serious period of time, the odds that not all your mail is getting delivered approaches and equals 100%. It's a sad reality but a true one unfortunately. If you okay with the fact that some percentage of your email will never make it to your destination (and in many cases you won't know that's happening) then it might be for you. Most people are not okay with that for obvious reasons.


> While you may not have liked the way it was said, this users reply is actually responding with correct information.

No, they are hand waving and being rude and insulting while delivering misinformation.


I'm calling out harmful misinformation. It makes me incredibly angry that it's 2023 and people are still peddling self-hosted email. That said, you have a point I've been unnecessarily rude in this thread and I should be more civil.


I've not been on any spam list in 10 years or more and only once in around 24 years (or more) of hosting my own email, and I do use my email accounts. That one time was 100% my goof, made a simple mistake, fixed it, was barely a hiccup.


> "You just cannot create another first-class node of this network.

Uh, yeah, you can, I did it again recently. I've done it at Oracle Cloud, AWS Lightsail, Linode, GoDaddy, and several ISP's before that. I could start right this minute with Digital Ocean, Vultr, etc. pick one and be up and running, delivering email to Gmail, Microsoft, etc. in a day.

Each time one of these articles is posted, I feel I must be some kind of email savant.


Here's the recipe: set up your own email server. Tweak configuration until eventually a test email from yourself to yourself lands in the inbox, then call it a day. Never actually measure your deliverability. Never investigate why you sometimes don't get replies to emails where you were expecting to get a reply to. In fact, just close your eyes and stick fingers in your ears. Then go on HN and talk about how easy it is to do this thing which you, totally, for real, really did do, like, for reals for reals.


Measuring deliverability is really not that hard. You do the following:

- Setup blacklist monitoring (e.g. HetrixTools/MXToolbox)

- Check if you can email Gmail

- Check if you can email Hotmail

- Check if you can email Office365

- Check on Microsoft SNDS that you are not blocked: https://sendersupport.olc.protection.outlook.com/snds/

Gmail, Hotmail and Office365 are the largest email providers and also the most strict ones.

I have accounts on all of these providers so testing deliverability is trivial. You could argue that testing deliverability with one account is unreliable but in my experience it simply is not. Usually you send a couple emails across a week and if they all go through you are good.

If you are paranoid, you can ask if any of your friends have Office365/Hotmail and email them. They probably have company or university accounts on there.

My server has only been blocked once after all this, and ironically it was by another company that self-hosts their email...


> I have accounts on all of these providers so testing deliverability is trivial. You could argue that testing deliverability with one account is unreliable but in my experience it simply is not. Usually you send a couple emails across a week and if they all go through you are good.

No, it's not trivial, and no, it doesn't work like that. It's fairly easy to get a test email to your own gmail/hotmail/office365 account delivered. It doesn't mean your other emails get delivered.

For example, when I needed to send out a link for my wedding photographs to my wedding guests, I did exactly what you suggested: I sent a few test emails to Gmail and Hotmail test accounts that I had set up, confirmed that they delivered, and then proceeded to send the actual email which had a link and like 50 people in the BCC field. Guess what happened: all the Gmail accounts placed my email in the spam folder. So then I had to later send another email, using a different email provider, asking people to check their spam boxes for the link (this time not including the link in the body of the email with the hope that it would increase the chances of delivery).

Note that this particular anecdote is not of self hosting email, it was using Migadu (I was initially not trusting that they can deliver email properly, so I ran those tests like you suggested, concluded that they seem to be delivering email, and then my actual real email was not delivered).


I guess I simply cannot share your pain because I've never had these problems.

Actually I initially got placed in spam too in Gmail/Outlook but I always replied to my own emails and marked them as "Not Spam" and the problem just kinda disappeared?

When I initially setup my server I was paranoid (as you should be) and so I had the following protocol:

- Send all urgent email with Gmail

- Send non-urgent email with my own server. If I don't get a reply within 24h, send again with Gmail, explaining my email troubles. Usually the other party would happily confirm whether or not my email landed in their spam without me even asking.

It was never a cold-turkey migration to my self-hosted setup. It was more of a slow process over a couple months after I gained more and more trust in my server.


> Actually I initially got placed in spam too in Gmail/Outlook but I always replied to my own emails and marked them as "Not Spam" and the problem just kinda disappeared?

Did you continue to use those same Gmail/Outlook accounts for your deliverability testing? Because that would explain why you're not encountering any deliverability issues in your testing.

Try to create fresh Gmail and Outlook accounts and see if you can actually deliver to them.


I have multiple Gmail accounts (like 5+ lol) as well as my own Google Workspace account (for unlimited GDrive) so for deliverability to Gmail testing I just used that.

I had a couple Hotmail/Microsoft Live/Outlook accounts too.

But deliverability for me was always quite consistent. If it went to spam on one account, it would go to spam on all of them.

At this point I no longer do deliverability testing because I simply don't see the need to. No one has ever complained about my emails going to spam so the thought has never really crossed my mind.

I've also had a lot of trouble creating Outlook accounts for some reason. New accounts seem to get suspended real quick. My previous Outlook accounts have mostly all been suspended too. Not sure what's going on there...

EDIT: From the output of `dig` I can see my landlord uses Outlook. I just emailed him a couple days ago and he replied so I guess all is well?


> But deliverability for me was always quite consistent. If it went to spam on one account, it would go to spam on all of them.

You mentioned that your emails on Gmail and Outlook initially went to spam, but that you clicked "not spam" on those emails. This action should prevent your emails being flagged as spam for that particular receiver who clicked not spam. Obviously it can't work such that one person clicks "not spam" on an email, and then everything from that sender is whitelisted for everybody. Because then the spammers could just register a single email account, click "not spam" once, and then spam everybody. So my point is, if you test deliverability on email accounts where you have already whitelisted the sender, of course it's going to look like email is delivered.

> I just emailed him a couple days ago and he replied so I guess all is well?

Presumably you and your landlord have already emailed each other in the past, so any anti spam system should allow emails between those addresses to reach their destinations. It doesn't really prove anything.


> You mentioned that your emails on Gmail and Outlook initially went to spam, but that you clicked "not spam" on those emails. This action should prevent your emails being flagged as spam for that particular receiver who clicked not spam. Obviously it can't work such that one person clicks "not spam" on an email, and then everything from that sender is whitelisted for everybody. Because then the spammers could just register a single email account, click "not spam" once, and then spam everybody. So my point is, if you test deliverability on email accounts where you have already whitelisted the sender, of course it's going to look like email is delivered.

Obviously I'm aware that me clicking "Not Spam" would not instantly whitelist me in Microsoft's systems, but the hope is that it would influence the spam algorithm somewhat. It doesn't even whitelist me on that account because even after clicking "Not Spam", my emails still went to spam for some time.

> Presumably you and your landlord have already emailed each other in the past, so any anti spam system should allow emails between those addresses to reach their destinations.

Maybe, but my entire university runs on Office365 and I never had any trouble sending to friends and professors there (who have never sent me emails previously).

I'm not really sure what you are trying to say here? Should I have some sort of commercial deliverability monitoring system setup? Do you have insights into what "real email providers" use?


I guess I'm mainly trying to point out how difficult it is to estimate deliverability. So, rather than pointing to a solution, I'm merely trying to illustrate that the problem exists. There's so many people in this thread claiming that measuring deliverability is "trivial" or getting deliverability is easy ("just do X,Y, and Z"). And when you get down to it, it's mostly people saying "I sent an email to myself and it went through" or saying "nobody ever emailed me back saying they didn't get my email", and so I'm poking holes at those anecdotes. No larger point behind any of this.

Or, if I were to put a larger point behind it, I would say that getting good deliverability is hard and it doesn't make sense to try it unless you are delivering email on behalf of a large amount of people (rather than merely yourself).


> Here's the recipe

We read from different cookbooks and my reading comprehension is high. My sent email stats shows over 20K successfully sent emails just for my personal account, since 1999, when I started tracking it.


> My sent email stats shows over 20K successfully sent emails just for my personal account, since 1999, when I started tracking it.

Ok, since you claim to have measurements, let's hear what they are:

- What is the total number of emails sent?

- What is the number of emails bounced?

- What is the number of emails that landed in inbox (as opposed to spam)?

- What is the number of emails blackholed? (recipient confirms delivery, but email doesn't show anywhere, not even in spam folder)


It's like coffee. For every study that says A, there's one that contradicts it.

I feel like for every pro host-your-own-mail on HN there is one that contradicts it.


I went down an internet rathole on coffee grinders. (I realize you were probably referring to health benefits.)

After many hours of this over many months, I finally looked for a comparison of the actual taste of coffee from cheap grinders versus 10-20x more expensive burr grinders. (As opposed to comparisons of grinder technologies, which are everywhere.)

There are almost no published side-by-side taste comparisons; when I did find one, the cheap grinders had won!


Yes, I was referring to heatlh benefits. Where I live, it's a saying that coffee gives you this and that, but you'll live longer if whatever.

I can imagine, because why would taste matter for a coffee /s ?


You should write an article on how you’re doing it!


Sure, but how long would your new server keep working (i.e., delivering mail to GMail, etc)? Do you know some magic incantation for staying off the naughty lists?

Setting up a server is trivial, yes. Keeping it going is a never-ending treadmill of not really technical problems.


Do you know some magic incantation for staying off the naughty lists?

Not magic, but when I managed outgoing Postfix servers for a few companies I had to set rate limits for yahoo.com an a couple other domains to reduce concurrency or they would block one of the SNAT's for a while. It probably sounds tedious but it really wasn't. There were not many MX that were as strict as Yahoo. I never ran into issues with Gmail but I think they cut some slack for corporate IP addresses and domain names.

For my own personal email servers I never had issues because I never sent at a rate that anyone cared about. The closest I got to that was running a forum that would email when threads would get updated and people subscribed to them but my solution there was to suggest to the people on the forum not to do that.


> Sure, but how long would your new server keep working (i.e., delivering mail to GMail, etc)?

I can't see the future, if the big email providers who likely have some of their trolls posting in the comments ever decide to start choking out us personal email server runners, then it'd be game over. If things remain for the next 20 years assuming I live that long then deliverability would be 100% for the next 20 years.

> Do you know some magic incantation for staying off the naughty lists?

I don't spam, that and don't make a finger fumble edit like I did the one time in over 20 years and didn't check to make sure it was working correctly first.

Also, I use http://www.mxtoolbox.com/ to keep an eye on my server.


It looks like my intuition was right, it's not just the technical sector of the economy...it's the economy and the administration in charge.


Since they blamed lower sales forecasts in China, I presume you mean the administration of Xi Jinping?


If you're into cheap or free and regardless opinions on Oracle, their Always Free resources are unmatched.


People actually want to install Windows 11 when better alternatives like Ubuntu/Kubuntu/Fedora/PopOS!/etc. exist?


There's people who like to play PC games or use Excel natively.


You have to precise recent multiplayer games with anticheats because that's probably the last category of games firmly locked on windows.

For old games, you could even argue that wine has a better compatibility story than Windows itself nowadays with the amount of stuff they broke in Win10 and Win11.


Agreed, was going to reply to parent's comment but yours is spot on. Wine can run some of my old games great. Steam Proton runs all the newer stuff I've purchased (oops, sorry Gabe I meant to say "licensed"!). I don't do a lot of MP except for stuff that has a native Linux client like TF2 and CS2. Well, I do play a lot of NMS and sometimes play it in MP no issues on Steam Proton though their MP can get a bit funky sometimes. I play 7DTD Windows version, usually SP but it seems like I've played it in MP, maybe I'm recalling incorrectly.


See my comment below on games, as for Excel, I use the online version, works great on Chrome, Firefox and Edge on Kubuntu.


Unfortunately, a whole lot (the vast majority) of enterprise engineering software is Windows-only. Be it Patran/Nastran, most of the Autodesk lineup or lots of others.


> Games though can be a nightmare, a recent nvidia driver seems to have created stuttering across all games using X to varying degrees. (downgrading the driver fixes the issue)

On ubuntu (Kubuntu really), and as a gamer, I've seen no issues with nVidia on X, both native Linux and Steam Proton games rock on.


The games use X, but I'm using Wayland. Wayland native is fine, X games are not.


My past experiences with SuSE, Fedora, and Red Hat, RPM distributions, eventually experienced a corrupt repository that in each case took hours to fix. That's when I settled on Ubuntu in 2006 and haven't looked back.


I heard bad things about RPM before, but that was long ago. To be honest, I think by 2006 most issues were already over, and YUM was mostly on par with APT. Don't know what might happen in your specific case.


There were several phishing attempts from that domain, onmicrosoft.com, to my personal email account this past week.


Microsoft.com has been constantly trying to fool me into subscribing to their Office tools.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: