At 10:47 we published an updated internal URL containing additional data for new functionality. This change was reviewed, tested & validated in our staging environment without any problems. Once released, our internal alarming immediately notified that clients were affected. The change was immediately rolled back but took a few minutes to take effect.
At 11:03 service was confirmed to be back up running.
We take this very seriously and are investigating why the initial tests didn't correctly reflect the outcome. Until that is resolved we will not publish any updates at all.