NameNode HA using supervisord - hadoop

Question

I am getting an error while implementing Hadoop NameNode HA using supervisord
when I run the command without using supervisord, it works fine,
but supervisord seems to be having problems especially with zkfc
The namenode supervisord configuration currently used in NameNode is as follows:
[program:NameNode]
command=/home/deploy/hadoop/bin/hdfs namenode
user=deploy
[program:ResourceManager]
command=/home/deploy/hadoop/bin/yarn resourcemanager
user=deploy
[program:DFSZKFailoverController]
command=/home/deploy/hadoop/bin/hdfs zkfc
user=deploy
What I'm wondering is,
What is the difference between 'hdfs --daemon start {something}' and 'hdfs {something}'?
When I checked the hadoop/bin/hdfs code, I don't know what the difference is.
When the above two commands are executed with supervisord, when the daemon dies supervisord does not automatically failover because supervisord does not recognize the first command, but NameNode HA works when the daemon is alive
The second command is recognized by supervisord and when the daemon dies, supervisord automatically performs failover, but NameNode HA does not work.
It means when an Active NameNode dies, another NameNode does not automatically change to active. By the way, if I restart zkfc of all NameNodes by 'supervisorctl restart', Active NameNode is elected, but it is one-time
If NameNode uses 'hdfs namenode' and zkfc uses 'hdfs --deamon start zkfc', NameNode HA works, but fails over automatically when zkfc dies.
In conclusion, I'm curious as to the difference between the two commands and why zkfc doesn't work normally in the second command.