Well, that's not a clustering issue, the result will depend on what you do in *your* code. Or I don't understand your point.
Well I try to clearfy my question
lets see we need to have application running on server A and I am logging some activities on server A. Now to remove the fail-over issue I make cluster and deploy the application on server B.
Now my questions are:
When I will start the application does log will be generated on both servers (A and B) or only A.
If answer is only on A then during failover of A. How can I ensure that log be produce on a single file.
I hope I you understand my question.
Well, it is really up to you to use the clustering framework so that it fits your requirements: read the doco, especially the ReplicantManager and HAPartition part. You will easily find a way to implement a "flip-flap" behaviour where you are sure that only node works at a given time.