Symptom:

• there are more than ten scheduled services on intervals which would require them to be run at the same time
• initially because they were all created at random times they will not be trying to run at the same time, but after a restart of the server the scheduler kicks 'em all off at the same time. So common choices like 10, 30, 60 seconds will mean that the tasks end up attempting to run at the same time
• one or two of the tasks get stuck with next run time set to some impossibly high number and do not appear to run.

Cause:

the default settings only allow 10 threads to be used at any one time for the scheduler, more than ten threads are required for the first execution of the thread.

Solution(s):

• bump up the number of threads for use by the scheduler (in settings->resources). Default is 10. Will require a restart however.
  or
• you could stagger the frequency of the scheduled tasks (e.g. pick odd seconds to ensure that they're unlikely to correspond).