This feels like it's pushing the TOS for Github Actions a bit by using it for post deployment, live operations. Like the hourly health check and your dead letter queue scheduler.
It made me curious, so I looked at their TOS[1], and they seem somewhat permissive about things like this. Though vague about where/when you cross the line.
Like, this seems clear:
"should not be used for...any other activity unrelated to the production, testing, deployment, or publication of the software project associated with the repository where GitHub Actions are used."
But then:
"but a low benefit Action could be ok if it’s also low burden"
Interesting. I use Github Actions to tag incoming bugs with a "needs-triage" tag. That is not related to production, testing, deployment, or publication, so I guess I'm not in compliance.
Purelymail does look great for personal use. But, be aware it wouldn't be economical for something like a big mailing list.
Try the calculator at https://purelymail.com/advancedpricing and put in some large number of outbound email sends, like 2 million in a year, and the cost gets into the thousands.
Purelymail sent emails are $2.03/1000 plus $0.18/GB. Compare to SES, which is $0.10/1000 plus $0.12/GB.
The complexity of setting up forwarding with AWS Simple Email Service/handling incoming messages without using their WorkMail is why I moved off it for my next project. There is no real option for this, although Amazon does provide a tutorial for rolling your own using a combination of S3 and a Lambda Function. [0] You'd think instead of creating a tutorial basically asking users to copy and paste their code, they may just want to add it as a feature.
In fact, I find almost everything on AWS overcomplicated. I ended up moving to a combination of MailJet for outgoing transactional mails and Zoho for incoming emails. This combination also happens to be free while in development.
A fair amount of that seems fixable with better color choices and fonts/size. It's curious to me that cloudcraft hasn't improved it. Lightish orange on a white background, for example?
Very cool! My friends and I built https://m8l.me/ which is very similar and we've been using it for the past year. Instead of forwarding to email address like maildog, m8l.me forwards all emails to your server (or any server) as a POST request. If you don't receive many emails a month you could use the free AWS EC2 instance and store the emails in sqlite, thereby creating unlimited custom email addresses on your domain for free.
M8L looks great, nice work! Are you able to give any insight into the architecture? I'm always paranoid when putting free tools online that they're going to suddenly get a wave of users & cost me a ton.
I used to use it before now. Now, I run my own email forwarding service at https://hanami.run to expose a nice UI and a Rest API so I can `curl` my emails.
Interesting article, I'm not sure if it applies to my situation or not. I use ImprovMX to forward several domains to one of my other domains.
I think one thing the article doesn't really mention is that the domain sending the email is only one (potentially small) factor in classifying email as spam or not. Gmail has done just fine classifying different emails as spam that are being forwarded from the same domain. After all think of how much spam is sent from the gmail.com domain.
I run email for two companies. In both cases, everything from SES goes straight to quarantine and users never see it. I would just reject SES altogether but once in a while something legit comes in and someone asks me to release it. I see a massive amount of spam from SES and pretty much all email service providers.
I don't see it as a which provider issue. If you receive email you will receive spam from all over. On the receiving end, I treat providers differently based on what we get. I see a lot of phishing and advertising from SES but we use Amazon so I can't reject it. I almost exclusively see phishing attempts from SendGrid so I outright reject email from SendGrid. So many things and people use gmail that I can't block gmail but I do increase its spam score. I see phishing attempts from gmail accounts almost every day, I report them to Google. Its the phishing that worries me most, seems incredibly easy to sign up for a service like SendGrid and start phishing people. Not one of these email service providers can/will globally block recipients in its network - one is forced to play a game of whack-a-mole, which means the bad guys always win.
In terms of outbound email, I have a very Dilbert answer for you. The marketing guy picked Constant Contact and we use that. I have no idea how good deliverability is with CC vs anyone else. We use an opt-in email list, don't send a lot of communications, and immediately honor opt-outs.
I usually just use gsuite + aliases/groups, which includes spam filtering, web console and a lot more. Granted, it's ~$6/mon, doesn't scale to 20,000 email addresses and you have to be cool with Google.
I don’t understand how this is any better than flipping on email forwarding on most domain registrars. Or why you wouldn’t use plenty of other services for a few dollars a month
Taking Amazon SES to production can be a little tricky, I think I am talking to robots with templated messages. Maybe I don't have luck or didn't explain myself well enough.
I didn't have any trouble with getting them to enable production mode SES for my personal email domains. The only thing I can think of that might have helped me (besides making it clear I wasn't using this for marketing emails) was that my AWS account had been open and in good standing for quite a while before enabling SES.
It made me curious, so I looked at their TOS[1], and they seem somewhat permissive about things like this. Though vague about where/when you cross the line.
Like, this seems clear:
"should not be used for...any other activity unrelated to the production, testing, deployment, or publication of the software project associated with the repository where GitHub Actions are used."
But then:
"but a low benefit Action could be ok if it’s also low burden"
[1] https://docs.github.com/en/github/site-policy/github-terms-f...