This incident has been resolved.

An issue with the Redis cluster caused a spike in API error rates. The team worked on a fix, and it seems the system is now stable.