Option to retain acknowledgement through recovery

Hello,

We are using Icinga for disk usage monitoring and sometimes see a disk fluctuate a little bit around the warning mark. We will acknowledge the alert, but if the service recovers and then goes back to a warning stat, it loses the acknowledgement.

We tried using the "Sticky acknowledgement" option but it seems this only works for services going between Warn<->Crit state.

Using the "Downtime" option is not really preferred because that also prevents an alert when a service goes critical, which we do want.

Ideally the flow would be:

1. Service goes from OK -> WARN
2. We acknowledge the alert
3. If the service recovers and then goes back to WARN, acknowledgement stays, indefinitely or until expire time is reached.
4. If the service goes from WARN -> Crit remove the acknowledgement and re-alert.

This is because we will usually receive the alert, ACK it and set an expire time for a few days so people can fix it. If meanwhile it goes critical we would still like to know, but changes state between OK/NOT-OK for a bit it doesn´t really matter since we already know about the issue.

Start by tracing the existing acknowledgement, recovery, sticky-acknowledgement, and state-transition handling in the monitoring core. Compare the current Warn↔Crit behavior with the requested OK→Warn→OK→Warn and Warn→Crit flows; done means acknowledgements survive recovery until expiry but are removed when the service becomes critical.