At 17:18 UTC 18 May 2020, one of the databases backing our API began experiencing issues. Between 17:18 and 17:21 a small number of API requests were dropped (approximately 800 total).
We responded to the issue and between 17:21 and 17:40 we were able to rectify the core problem. By 17:47 we were processing at normal capacity. It took approximately one hour to work through the backlog of API requests that had accumulated. Triggered messages were delayed during this period. No data was lost.
Aside from the ~800 dropped requests, no data or emails were lost. This issue was caused by a regression introduced with a recent infrastructure change. We are reviewing our tests and QA moving forward.
Thanks. If you have any questions, please write to us at firstname.lastname@example.org
— we're happy to answer them.