I recently posted about an incident that occured with a third party app (in this case it was MAKE but could have been any other system), where a very unusual and unforeseen condition occurred that caused that system to constantly trigger MAKE to fire off actions on Tadabase. As a consequence it used up 10,000’s of api calls on Tadabase and Make until they both run out of daily credit. Had those daily limits not been in place it would it have run indefinitely. Since each individual trigger was valid in it’s own right MAKE didn’t see it as an a problem and the scenario just looped.
Having had time to reflect, I’ve realised I need to do a lot more in an attempt to error handle and identify those conditions however it’s not very easy to achieve as you’d need to historically keep track and check if the same action was being re-triggered.
I wondered if Tadabase would be open to adding some safe-guards to help trap such errors. My system wont ever send more than a 1000 api calls in an hour so if we were able to set app thresholds and say a way to “trip the switch” like a fuse if we had a run of events would give me some piece of mind.
I know ultimately it’s down to writing better code and better error handling but since each instruction was valid it’s not easy to manage.
Also I wondered if Tadabase could provide an api call that allowed us to halt further API so if we did spot suspicious activity we could shutdown Tadabase from accepting more API calls and we could alerts ourselves of the event so we can investigate further.