There is no concept of 'epsilon' in 2-Node cluster. In short, neither node holds epsilon; instead, both nodes are continuously polled to ensure that if one node fails, the other has full read-write access to data, as well as access to logical interfaces and management functions. Hence, by-default the "surviving node" automatically becomes the RDB Master.
It doesn't matter which Node you want to go first. Just make sure that you perform the 'halt' from the Service Processor console.
1) Note the IP of the SP for both Nodes from the cluster shell. ::>sp show
2) Then, login to both Nodes via SP IP and then perform the 'halt' command as mentioned in the article. Once the OS is gracefully shutdown and returns to loader prompt, you are safe to physically power-off the controller and the disk-shelfs.
Are far as turning it back on it tells me to plug in and then power on all core switches. Is that referring to the switches that the netapp is plugged into or is this referring to to just the switches that connect the nodes. Because we only have the two controllers, we have a switchless cluster.