At 16:04 UTC, September 8th, we made an enhancement to our routing protocol for compatibility with a new operating mode, which caused the current routing system to fail. The change has since been rolled back, and the issue identified to be resolved. As of 16:48 UTC, September 8th, full functionality has been restored.
In response to this we will be augmenting our monitoring processes for new version rollouts, ensuring we can react to these kinds of changes more quickly.