Web site down for 20 minutes

During a scheduled upgrade today, an error caused TLS certificates to not be deployed correctly.

The incident happened at approximately 10:10 UTC and we were notified at 10:25 about the problem. At 10:30 UTC the problem had been fixed and all services were functioning again.

The problem was only on one of two load balancers, so we can assume that approx. 50% of requests had errors during the incident.