Cloud providers (Azure, AWS) let you set up many servers in a scaling group, but most of the servers will not be turned on until they are needed. They will be turned on and off based on server load. This leads to scenarios where (for example) 3 of 7 servers are intentionally turned off, as they are not needed. In this case, I do not want Shinken to tell me that the servers are turned off, because I want them to be turned off if they are not needed.

However, if a server is turned on when the service gets busy, I now want Shinken to tell me if that server is not working like it should.

Can the discovery module work with these kinds of situations?

If not, does anyone have any recommendations on how to approach solving this problem?