Shut down crs on RAC, including two nodes(dukerac01/02).

--> log on to the node1 and issue the following commands
 srvctl stop database -d ORDB
 srvctl stop asm -n dukerac01
 srvctl stop asm -n dukerac02
--> run "crs_stat -t" to check resources' status again.

then:
 srvctl stop nodeapps -n dukerac01
 srvctl stop nodeapps -n dukerac02
--> run "crs_stat -t", and all resources' status should show "OFFLINE"

then:
 sudo crsctl stop crs

--> ssh to the dukerac02 server, issue the following command
  crsctl check crs
--they should show "OFFLINE" healthy

then:

  sudo crsctl stop crs
--> last step, run "ps -ef" to see if there are other resources related oracle exist.

Starting crs on RAC, of course, just take the opposite steps above.