I have an aws emr cluster running an spark direct streaming application. When for some reason a core instance fails (I tried this by manually terminting it, but in the futur
