Sometimes in on-premise deployments there can be intermittent network drops between pods. This can cause issues with the pods that communicate with redis (backend-listener, system-reque, etc).

The issue presents itself as either timeouts or connection reset errors.

Connection timed out (Redis::TimeoutError)

Redis::ConnectionError - Connection lost (ECONNRESET)

It would be useful if the clients that communicate with Redis retried the connection.

When present, this issue will also cause a significant amount Admin API calls to fail completely due to the redis connection error. If the connection could be retried before failure it may prevent this.