The built-in low-level discovery support is great for discovering filesystems, network interfaces, CPUs, and other entities. But what if we have some custom software that we would like to discover components with or perhaps are running an older Zabbix agent on some system that does not support a particular type of discovery yet? The great thing about LLD is that it is very easy to extend with our own discovery rules. Let's take a look at two examples:
- Reimplementing CPU discovery on Linux
- Discovering MySQL databases
An LLD rule never returns item values. It discovers entities that allow the creation of items from prototypes. Items receive values from agents, SNMP devices, using zabbix_sender, or any of the other data collection methods.