While improving the inner workings of our application we accidentally caused a failure which resulted in a downtime of 22 minutes. This was resolved and our engineers are fixing the root cause to prevent this from happening again.