Host with SSH checks: Availability and Dependencies

[MickeyPM]

Hi,

We have a few sites that won't allow the agents to be installed, but we have setup ssh keys between the Zabbix server and their servers.

Is there a simple way to run a check item to do a local ping from the Zabbix server and set the status of the Server to Available ?

Also base on this ?
From what I understand items are checked regardless of the status of the host availability!?!?

Are there some sort of item (or even template) dependencies that can be used to stop child checks if the parent check fails ?
or a way to add a check in an item to stop the actual item check from doing the remote call (which will timeout), if the server is unavailable ?

Parent ( ping )
--> Child 1 (memory)
--> Child 2 (cpu)
--> Child 3 (load)
--> Child 4 (etc.)

[tim.mooney]

Zabbix supports dependencies between triggers, the functions that decide if a threshold you've set has been exceeded. You could make a simple check "ping" trigger the base dependency for a host, and make other triggers depend on that. If the ping check threshold detects a problem, then the other triggers will not fire.
This is detailed in the trigger dependencies documentation.

Note that Zabbix will still continue to attempt to gather item data even when a "base" trigger indicates there's a problem. It's the dependent triggers that are suppressed, not the (attempted) data collection.

Trigger dependencies, including network topology dependencies, are a good way to avoid "alert storms" when e.g. a core network device is temporarily offline.

[MickeyPM]

Thanks for the feedback, but it defeats the purpose I am trying to achieve.
The triggers are too late, I want to stop the items from running.

(I experienced the issue on the monitoring system that we are busy replacing.)

I have around 50 item checks and the amount of load and processing created on the monitoring server when all these item checks are taking too long waiting for the ssh checks to time out due to the server not being available. And multiplied when a few sites go down, just kills the monitoring server.

I pressure the agent interface must have something like that built in whereby it’s only run the items when the agent is active.
Is there not something similar for the SSH agentless option?

[dimir]

If you use passive checks (which SSH checks are) then host availability will be based on those. When host becomes unavailable it stops being checked. In order for a host to become unavailable one or more of its items must subsequently return network errors. When host becomes unavailable it starts being monitored by unreachable poller processes. When network errors disappear the unreachable pollers set it is back to available and the checks continue. More on that here: https://www.zabbix.com/documentation...unreachability