Very unfortunate timing there. We had a cache server wedge, and the failover mechanism didn't detect it quickly enough. We've diverted traffic manually and are working on bringing the bad region back into the mix after it went down.

Things should be working again now, at least.