CMK version: 2.4p39/2.5p15
OS version: Linux
When it comes to delay a service notification, then the Nagios core and the CMC show different behaviour.
In short: the CMC does it right 
When service goes OK → WARN, then both cores correctly process a service notification delay rule as set.
If the same service goes WARN → CRIT, then the CMC correctly schedules a new delay for the CRIT message. The Nagios core does not, it will send out a notification immediately.
Furthermore, if the time span that is used for all check attempts is higher than the Service delay rule (f.ex. 60s interval with 10 attempts = 10minutes and Delay rule = 5minutes) the Nagios Core will ignore the delay rule and send out a notification as soon as the hard state with the 10th check attempt is triggered.
The CMC correctly schedules the delay
I know that the Nagios core will not be patched but maybe someone has a good idea to overcome this in Community Edition Checkmk?
Best
Tom