1.
In the official template "OS processes by Zabbix agent", the Processes discovery rule (key custom.proc.discovery, dependent on "Get process summary") has Disable lost resources = Immediately by default.

This setting stops the template from doing its main job of alerting when a monitored process gets terminated.

What happens:

1. A monitored process gets terminated.
2. The process is no longer in the proc.get summary output, so the next discovery run marks it as a lost resource.
3. Because of "Disable lost resources = Immediately", the discovered items and the "process is not running" trigger are disabled at once.
4. The problem either never fires or fires and then disappears a few seconds later. Nobody gets alerted about the stopped process.

Expected behavior:
When a monitored process stops, its items and triggers should stay enabled. The "process is not running" problem should stay open until the process starts again.

Request:
This must be reworked - change the default in the official template, preferably to disable After with a reasonable period.

2. The LLD rule gets performed every time when its master item gets polled, i.e. each 1 minute.
This is against generic best approaches approach to not run LLD rules too often, as its a bit resources consuming routine.
Normally there should be 2 preprocessing steps - JsonPath to take only $.name and $.vsize and then discard unchanged with heartbeat.