Starting around 6:13pm EDT on Thursday, an unusually high spike of messages caused our message queueing services to fail. Help Scout services (web app, mobile API, and help desk API ) were down for roughly 18 minutes, and services returned to normal around 6:31pm EDT.
We're working to separate our message queue cluster to spread any unusual spikes like this across several clusters. Going forward, we're also adding new alerts, as well as additional error logging around our internal APIs.