When performing maintenance to a cluster, as in adding an SQL Server virtual to a node or removing it from a node, setup may fail for a number of reasons. And once something went wrong, things keep getting wrong, even if you verified the procedure and you corrected the cause of the error.
Well, the procedure didn't tell you everything. First of all, you should only be logged in to the node where you are running the setup. Logoff from all other nodes. The reason is that SQL Server setup will create a Windows Scheduled task on the other nodes of the cluster for you, under your account. Having an interactive session (you being logged in) and logging in as a batch job (the setup task) with the same account, may lead to unexpected events (and thus setup failing). Obviously Task Scheduler should be running to . To be honest, Microsoft does tell you about the scheduler and the logging off in a KB article (let's be nice and add some community content).
Where it does get mean in the KB article, is when setup keeps failing (and it most likely will); restart all remote nodes!!! The most likely reason for the setup failing this time, is a running setup.exe from a previous failed attempt. This is what may cause the access denied for the scheduled task (or check that your account is local admin on every node and has the login as a service permission). Here is how to avoid having to reboot;