SDLCLI healthcheck shows wrong (NOK) status after Kubernetes dbaas statefulsets has been scaled in and out, because Redis sentinel has a bug what leaves extra sentinel entry with zero port number in configuration. This Ghost entry is also visible in Redis client "sentinel sentinels dbaasmaster" as an extra sentinel with flags field indicating an error (s_down,sentinel,disconnected). There has been reported already a Redis server bug report (Remove duplicate zero-port sentinels (#9240)) and that has been fixed but fix is not available yet in any Redis server releases what we use.
As a fix for this Jira issue, until Redis server fix is available change SDLCLI healthcheck implementation to ignore extra Redis sentinel entries with zero port number from Redis client "sentinel sentinels dbaasmaster" output.