In our efforts to continuously improve our product we deployed a change that had an unforeseen negative performance impact. Unfortunately, this change caused downtime in one of our European clusters.
The length of the outage was due to the fact the newly introduced feature has been deployed multiple hours earlier, making it harder to pinpoint as the root cause of the issue affecting only one cluster.
After we found the root cause we reverted the change and the main database required some time to recover.
We sincerely apologize for the inconvenience. To avoid this kind of situation in the future, we will rely more and more on progressive rollouts and performance testing when releasing new features.