That would complete the thing. Each device will pull the config from server once in hour, check the md5 sum, and if differs perform a restart.
The centralized PRTG monitor will fetch details of all devices including the temperature and throw an alarm with location of any device above usual.
So far, everything behaves normally even with most extreme loads, but I guess the real test is going to be during summer.
Thanks for the params, very useful both for monitoring and throttling! Very helpful.