The fact that this doesn't exist after 8-years is mind bogglingly absurd

We just saw another batch of 25 instances this morning where we were forced to send potentially duplicated emails because of connection interruptions. Wanted to quickly renew the call for this feature. Would provide us devs with peace of mind and allow us to avoid confusing our customers with duplicated payment receipts or other transactional emails.

Would be good to support this please 🙏

We've seen connection errors that resulted in re-attempted (and in many cases duplicated) emails almost 50 times in the last month. Some of these are minor, but others are potentially alarming (e.g. payment receipts, where a user might worry they'd been charged multiple time).

We need this feature too.

This feature is essential to increasing the resiliency of our integration with the Mailgun sending API without risking duplicate sends to our customers. Adding the idempotency key would have allowed our system to deliver uninterrupted service to the majority of our customers despite interruptions to the API.

Show custom headers in the events api to help identify emails sent.

Sometimes mailgun will trigger a 5XX error for some emails sent through the api but still receive and send that email. I have verified this in the logs. I have a customer header I use in my emails X-MS-ID=someid I use to identify emails. I can use the search api to find and identify emails which I don't have an id for and prevent sending duplicate emails resulting in user complaints.

https://documentation.mailgun.com/en/latest/api-events.html#event-structure

Sometimes when sending with the api we will get 500 errors. When we go to look our end requeues it for a retry and tries again but mailgun actually received and sent the email successfully already. We end up sending twice to the same email with no solid way to determine if an email was already sent. The current api only lets you lookup with an id we do not have because we got a 500 error.

Would be still important to us - makes our systems more fault resistant

Any progress on this?