This Knowledgebase article provides information about why Windows Time Service should not run when Neverfail Heartbeat is running.
If the time on a Neverfail Heartbeat server is corrected backwards (by a few minutes) then that server will stop issuing heartbeats, and the connected server will get heartbeats missed. If this happens to the active server, the passive will get heartbeats missed and failover, causing a split-brain syndrome, as both servers are still connected. At that point, application protection is lost and manual intervention is required to reconfigure the Neverfail Heartbeat server pair and restart replication.
Server clocks are not synchronized. This issue has been seen on a few separate occasions and on each occasion, it was triggered when the Windows Time Service adjusted the clock on the active server backwards.
The origin of the issue appears to be a known issue in Java (4239522, linked below) that Sun currently has no plans to fix; therefore, the only option is to ensure that the Windows Time Service is not running on any server in a Neverfail Heartbeat Cluster.
Sun currently has no plans to fix this; the only option is to ensure that the Windows Time Service is not running on either the Primary or Secondary server in a Neverfail Heartbeat pair.
Sun Developer Network Bug Database, Bug 4239522 — Timer seems to stop when the computer systemdate is turned back
Knowledgebase Article #53 — Server Clocks are Not Synchronized
Knowledgebase Article #1551 — How to Create a Task to Synchronize the Time of the Passive Server with the Active Server When Neverfail Heartbeat V5.3 and Later is Installed