The Xumm API platform & backend was unable to reach the Redis pub/sub cluster, responsible for informing clients & API consumers about paload status updates. While the Redis cluster was onl offline for a brief period, when it came back, it was overloaded with retries, resulting in the cluster going down again.
The backend has been updated to cancel retry requests sooner (to prevent overloading the Redis cluster in case of downtime) & this scenario will be added to monitoring.