It is observed alert definitions which are configured with the default 1 wait / 1 cancel cycle are triggered on the 2nd polling cycle (collecting cycle) instead of the expected 1 polling cycle.
The configured alert definition is configured with a symptom that is triggered on the Power Off on objects such as Virtual Machines.
Aria Operations 8.x
If a object is monitored by multiple adapter instances it may cause alerts to trigger later than expected (maximum of a 5 minute gap) as the object resource is shared between multiple adapter instances.
This is expected behaviour and is a known issue.
To check how many and which adapter instances are monitoring a specific object, navigate to the "Details" page on the object and the adapter instances will be listed against the object.