We had some problems with page serving flow. The cause was communication issues with the database which resulted in some pages having intermittent 404s. The issue started around 1 PM UTC - full stabilization around 8 PM; the worst moment being between 2:30 PM and 4 PM UTC.
We're still analyzing logs. We already found many clues which can give us concrete implementation ideas to improve our system so that this won't happen again in the future.
We apologize if you were among the users that experienced this issue, and we thank you for your understanding!