No action is required. VCS is taking action on the FAULTED resource.
To analyze the reason of the unexpected offline, check the agent log or the engine log.
If you do not want VCS to take any action when the resource is reported as OFFLINE by the Monitor entry point for certain times, raise the value of the ToleranceLimit. The ToleranceLimit attribute is the number of times the Monitor entry point should return OFFLINE before it declares the resource FAULTED. Its default value is 0. Note that a large value for the ToleranceLimit attribute can delay detection of a genuinely faulted resource.
For more information about ToleranceLimit, refer to the Veritas Cluster Server Administrator's Guide.