Look at
http://blogs.msdn.com/clustering/archive/2008/05/10/8483427.aspx - there is a list of useful documents related to clustering.
Your SQL has to move to secondary node! It's why you created a cluster... If you like to control entire process manually you can take your SQL resource offline (in failover cluster management console) then do your normal maintenance task and then bring it online again (after ensuring it is located on a right node).
Failback options can be set in group properties. failback is disabled by default because failing back generates next service availability break.