I have created a Custom Alert like this:
Then I have done som testing:
In this test case I have a total of 22 failed requests (7+8+1+5+1) but the problem raised by the custom alert looks like this:
Could someone please explain how the value of 4 is calculated?
sum in this case (in my opinion) is sum number of occurences during 2 minute period during 4 minutes interval. you cannot exactly say when the 2 minute period was started, it can be anytime at beginning of error interval, where it's possible that sum of errors per 2 minutes were 4..
sum is the metric aggregation you selected for each minute slot. Each metric is (in highest resolution) represented by minute slots that offer min/max/sum and count values. The sum delivers the sum of all errors that were recorded in that minute.
So if you define a custom alert on sum on a sliding window of '2' out of '4' minutes. this means that in a 4 minute sliding window at least 2 minute slots have to deliver a sum of errors higher than your threshold of '0'.
The value of '4' should be the worst severity value that breached the threshold, but this seems a bit low here in your example.
Does your last sentence indicate that I should open a support ticket? Because if Dynatrace should behave as you described, then the value should be '8'.
I tried changing my custom alert setting to 'For 1 minutes during any 4 minute period' and then only generate 1 http 500 error. The resulting alert was:
And in this example I would expect the sum of errors to be 1