UI occasionally seems to be showing backend down message which self resolves in a couple of seconds. It is however annoying for users who are frequently disturbed by this behavior.
The underlying issue seems to originate from the 502 Bad Gateway error which is not something that we can control from within CDAP.
As a solution for now increasing the polling interval for backend status and waiting for 3 retries before we show the backend down message. Essentially if the backend is down for 30 seconds we will show this message in the UI.