Description of problem:
A restart of beaker-watchdog will cause a spike in xmlrpc calls for any active recipes.  This is because there is no state saved in beaker-watchdog and it will re-send log files from the beginning since it doesn't know what it sent before.

Version-Release number of selected component (if applicable):
0.5.56

How reproducible:
Everytime the service is restarted

Expected results:
One solution would be to ask the scheduler the size of the file upon initialization,  this would introduce a small overhead for every Watched file, but would mean restarting the watchdog would not re-send already transferred data.