Need help to document the basic guidelines/best practices  for creating a CRD

 I was going through available Red Hat/Upstream documentation to find what are the best practices and guidelines to be followed while creating a CRD in openshift. i have a customer who is insisting for the Red Hat documentation for the same.

Please find customer comment below.

~~~
The problem of the available documentations is that there are basic examples but there are no information about the issues which can be created if CRDs are used in a wrong way. The basic guidelines for creating CRD are missing! 
A developer just reads this pages and is not informed about the issues he can create. Also he does not know that he is creating objects in ETCD directly because he does not care how OpenShift is handling this internaly. 
People can create performance issues in the whole cluster by creating CRD's.
We also already had the situation that wrong version handling of CRD and CR killed the apiserver of the whole cluster because someone "just replaced this CRD". 
There is a lot of what can wrong here and Red Hat should provide information to avoid performance problems or complete cluster outages. 
~~~

 Not Applicable

 Not Applicable
  

 Not Applicable