jdamick
commented
12 years ago Tasks:
- Define an interface to properly register resource callbacks (assign some callback ID)
- Create validations to ensure that the resource formatted correctly and is accessible during registration
- Store the resource and threshold condition
- Create a filter on metric writes to trigger callbacks that will be service asynchronously by some metrics service instance
- Distributed Job scheduling of sort will be required
- Needs to retry over X period
- Define an interface to reregister callbacks (by ID)
(this is pretty big)
As a metrics consumer I want to be able to specify a callback resource to be triggered when a conditional threshold for a metric key's value is met so that I can send alarms or notifications
Given that I have a callback resource When I register the callback Then I must specify condition for a metric key's value to trigger on And my resource must be accessible by the Metrics Service
Given that my resource is not accessible at trigger time When the service attempts to access my resource Then the service will retry up to 3* times over a 10* minute period (**just random values)
Given that I have a callback registered When my registered metric key's value exceeds the threshold value I've specified Then I will receive a POST* to that resource given the key & value metric
Give that I have registered a callback resource When I deregister my callback Then I will no logger receive POSTs to my resource
* Again, I assumed REST JSON style interface..