We experienced a global outage due to an undiscovered software bug that surfaced on June 8 when it was triggered by a valid customer configuration change. Here's a rundown of what happened, why, and what we're doing about it.