You should put an OEM Blackout in place for any planned interruption of Oracle service
This assumes you are using some form of Change Management system and that an environment variable called CHANGE is explicitly set to the ChangeID that this Blackout will be set to support. For example:
export CHANGE=CM20999999
EMAGENT_HOME is set in the environment. Ideally this should be in the .profile for the 'oracle' user. For example:
export EMAGENT_HOME=/u01/app/em12c/agent_13.1.0.0.0 # Database Servers
All commands should run as 'oracle' when on Database Servers (and 'oms' on an OMS)
AGENT_HOME is likely to be different on the OMS. Ideally this should be in the .profile for the 'oms' user. For example:
export EMAGENT_HOME=/u01/app/agent12c/agent_13.1.0.0.0/bin # OMS
cd $EMAGENT_HOME/bin
./emctl start blackout ${CHANGE} -nodeLevel
./emctl start blackout ${CHANGE} ${ORACLE_SID}:oracle_database -d 00:10
cd $EMAGENT_HOME/bin
./emctl stop blackout ${CHANGE}
cd $EMAGENT_HOME/bin
./emctl status blackout