Incorrect SLA calculation for parent services

For example, we have IT services with the next struct:
serviceid = 1 is parent for serviceids 2 and 3. Type of calculation for Serviceid = 1 is "Problem if at least one child is in PROBLEM". Serviceids 2 and 3 have different triggers.

Sometimes when triggers from serviceids 2 and 3 are changed for 1-3 seconds and they are processed by different syncers, then service_alarm for serviced = 1 is missed.