Prospect was test dynatrace and kill application pool process but no problem was started because of that ? Is there any explain for that?
Also prospect ask what is the amount of time that dynatrace decide that the problem is starting and the alert is fired ?
Yos @gil g.
Solved! Go to Solution.
Most of the problems the AI Engine detects aren't as simple as "process is down, fire alert" but instead look to see what transactions and requests are failing for real users and then attempt to isolate what caused these failures or slow downs (based on baselines gathered over time). So if the process is down but not a significant number of actions are actually failing then it is unlikely a problem would be reported out of the box. The idea behind this is with cloud based architectures and microservices a process or host goes down but either this is expected or it recovers automatically so you wouldn't want an alert unless users end up being affected. This is why it is important to make sure your deployment is being "fed" enough data so that the AI can learn and have good data to work with.
Of course if you would like more sensitive alerts these can be configured. In the case you're testing you could set up alerts at the process group level that would create a problem if fewer than X instances of that process are active. You can also select "key requests" which will create problems more quickly if they experience problems (e.g. a checkout transaction on an eCommerce site). More info on problem sensitivity adjustment here: