We have recovered operation of our Cassandra clusters and are seeing standard API latencies.

We identified a pattern of access causing some underlying strain on one of our Cassandra clusters, leading to bad latencies for message sending/retrieval. We've addressed this one instance and are now working to understand what happened to the circuit breakers that are supposed to (and have historically) caught & prevented this kind of degradation.

At the moment, though, things look like they're recovered and should be fine. We are continuing to monitor as we work through understanding the root cause.

We are aware of an issue sending messages. The team is online and investigating.