[Security and Access Control] /etc/init.d/ego status overiding default behavior - not according to standard, breaks monitoring and automation tool chain

The init.d script being rolled out with Symphony 7.1.0 is the same across all compute and management hosts. Running the script in its intended way with the status flag always results in the same message:

-bash-4.1$ sudo service ego status
ego service is stopped.

Which is not true, the services (lim, pem) are up and running. The script tries to also run "egosh ego info", which can only be run when the platform.profile is sourced and the user executing the script is logged in to the the cluster. This is never the case for any of the use cases described below, only our application admins are logged in to the cluster when running administrative tasks. No other tools or admins should be logged in.

There's 2 major problems with this approach:

1) the init system is a system to bootstrap processes and check for their LOCAL well-being, by the means of providing a status routine that checks for the processes pid. In no case shall the status flag require anything above the local host's stack, like network or application level functionality. It's crucial to limit the functionality and exit return codes to just local services. Custom functionality can be added by custom flags that do not override the core flags. In this case, this is not even necessary, as "egosh ego info" would only be run by a cluster admin, not a regular local admin who would be interested in the status of the local processes only. Checking for cluster info must not be part of the init system.

2) Ignoring the impact of this puts adoption at stake, and increases support effort on IBM's side due to the false error messages.

We provided a fix (3 lines of code) months ago, yet are still debating about why it's a good idea to follow standards and allow tools to communicate with each other. The proposed solution from IBM support and development side breaks other customers' use case if they really rely on checking the cluster status by the means of an init.d script (which they shouldn't do in the first place). In addition, that solution also mixes the local process status with the clusterstatus in one go, again.

No progress so far. See PMR for details.