Between 5 and 6pm PT on 3/30, Redis underwent a "host migration." We were not alerted to this in advance, and it resulted in 5,682 failed requests. We're currently working with Redis to create redundancy that will help us avoid these kinds of outages in the future.