See www.zabbix.com for the official Zabbix site.
New concept for UNKNOWN events
Owner: Alexei, Sasha
A monitoring system should produce two types of events: problem and recovery. Zabbix currently generates also 'unknown' events, which is quite confusing to Zabbix users. Why shall we tell to the world that Zabbix is unable to execute certain checks or recalculate trigger expression? It should be considered as an internal event, i.e. event not exposed to end users or external systems.
It will also bring a number of short- and long-term benefits:
- No confusion related to the unknown events. List of events will be much cleaner: a single OK instead of OK->UNKNOWN->OK.
- Same logic (no unknowns) for events and IT Services
- Only Zabbix server will produce new events, GUI won't do it anymore.
- Performance of GUI for host, trigger and item related operation will be improved
- Zabbix Server will not generate unknown events on startup. Server startup time will be significantly improved.
- Zabbix Server will not use IDS table for generation of new events, which will lead to much better performance.
- No deadlocks related to IDS (see also 1.)
Two types of trigger based events will be generated: PROBLEM and OK. It means that only significant events will be produced.
A new type of events will be introduced: internal events. Internal events will be used to notify Zabbix administrators about configuration related issues, for example, unable to calculate trigger value, item goes to not supported state, etc. These events will be visible to administrators only.
Specification (PART 1. Removal of unknown events)
Specification (PART 2. Internal events)