It looks like HDFS is started on all your nodes but YARN only started on the master. If you’re using Ambari then you can try to start the YARN processes on each host through the web interface, otherwise you can try starting the NodeManager manually on each slave machine.
It doesn’t look like Hadoop 2.3.0 is officially part of any of the HDP releases, but you can probably refer to the HDP 2.1.5 manual installation book.
Basically on each NodeManager node, run as the yarn user: yarn-daemon.sh start nodemanager
You’ll probably need to specify the path to the shell script, and you might need to specify HADOOP_LIBEXEC_DIR and –config $HADOOP_CONF_DIR if they’ve been changed from defaults.