Configurable retry window longer than 8 hours for temporary delivery failures
Hi Mailgun Team,
We would like to request an option to extend the retry window for messages that hit temporary (4xx) delivery failures. Today the window is fixed at 8 hours.
Background On Thursday, 11:56 PM, Microsoft 365 / Exchange Online had a global outage (Microsoft incident MO1550821). Exchange Online rejected our messages with temporary errors, such as 421 4.3.2 and 451 4.7.0, for about 17 hours.
Recovery for our domain came around Friday, 5:30 PM.
Mailgun retried correctly, but the outage lasted longer than the 8-hour retry window. As a result, 11,377 messages expired with "Too old": 7,679 to our own domain and 3,698 to clients. None of these messages had a problem of their own.
They only needed more time. Our teams then had to identify, de-duplicate and manually re-send them, and some time-sensitive client notifications were lost.
Based on our logs, a 24-hour retry window would have kept every one of these messages in the queue until Exchange Online recovered. A 16-hour window would have saved about 68% of them. Microsoft extended its own email timeout to 48 hours during this incident for the same reason.
What we are asking for
A configurable maximum retry duration at account or domain level, for example up to 24 hours or more, instead of a fixed 8 hours.
The ability to extend the retry window for messages already queued during an active incident, through the API or control panel, or by request to support.
An option to pause and later release delivery to a specific destination, such as Exchange Online-hosted recipients, during a provider outage.
Alerts or webhooks when the queue grows unusually large or when queued messages are close to expiring.
An easier way to bulk re-send "Too old" messages from stored copies.
Even just the first item, a configurable window up to 24 hours, would have prevented all message loss in this incident.
Mailgun Users