zabbix agent: set panic on not existed provider optionally

[ipslot]

Add use case support: On same host (zabbix node) running more than one agent applications and sending different data to zabbix (I mean different metric providers).

[michaelquigley]

Thanks for the contribution.

I'm not sure I understand the intent behind it. The Zabbix protocol is designed to either return a metric value or return ZBX_NOTSUPPORTED. What is the reason for throwing an exception here?

[ipslot]

In reality, I ran into a problem. On one zabbix node there are two applications with built-in agents. The first application sends the key "application.A.uptime" the second sends the values of key "application.В.uptime". Before the fixes, both applications send both keys, although in each of them only one metric provider is implemented. This fix allows the agent not to send a ZBX_NOTSUPPORTED error if a profiler is not specified for this metric. This behavior is optional and the default behavior is as before.

[michaelquigley]

So rather than trying to change the behavior of the agent, is there a way that we can alter your usage scenario so that the change is not required?

It concerns me to make the agent work in a way that isn't "correct", even if that functionality is made optional.

I'd like to better understand your scenario, so that maybe we can find a solution that will work for everyone?