My best guess would be a datacollection problem for your ESX servers, normaly datacollection is sheduled bi weekly, so the events shoild only happen one every 2 weeks arround the time the job runs, there also is a log for this job you can look at or trigger the datacollection from the menu under option when you selected the ESX servers that you have these problems with.
Sometimes an identify gives more insight as the logging for that job is more complete. If you see errors in the log you could post them here for troubleshooting.
