The outage was caused by planned maintenance of a database, which had unforeseen spill-over effects.

We’re pursuing various mitigation strategies to make sure it doesn’t happen again, chief among them is improving our application level caching to reduce the dependency from the platform to the databases.