Elevated send failure rates

Incident Report for Superhuman

Postmortem

We deployed a change to simplify the HTTP client that we used. Due to a difference in configuration in production vs. development and test, we did not catch that this would break in production.

We've now updated the tests to run the code in a more production-like way to avoid this problem in future.

Posted Oct 12, 2017 - 15:10 PDT

Resolved

All affected sends are now sent.
Posted Oct 12, 2017 - 15:06 PDT

Identified

A bad deploy is causing some sends to fail when they shouldn't.

We've rolled this back and are working to manually retry the failed sends.
Posted Oct 12, 2017 - 15:04 PDT