To determine where the problem lies, it is important to review the IntroscopeEnterpriseManager.log file in the <EM_HOME>/logs folder of the MOM/CDV and look at the warnings. If they only mention a single collector, then the problem is most likely on that single collector. If they mention multiple collectors, then the most likely cause is a problem on the MOM.
The network time service for the problematic server needs to be checked to verify that it is running, and that it is configured properly. A quick way to check on a Linux system is to run "ntpq -p" - it should run successfully, and return a table such as this:
remote refid st t when poll reach delay offset jitter
+220.127.116.11 10.10.10.10 2 u 135 1024 377 0.000 0.000 0.000
If the returned delay/offset/jitter are 0 values (as above), then it indicates that there is a problem as these should always return a non-0 value. You or the system administrator will need to verify that the NTP service is properly configured and working.