Server availability check from multiple locations

[Makihiko]

We have a single Zabbix server which monitors around 80 servers distributed across multiple continents. We have a trigger that alerts us when a server is unavailable (based on zabbix[host,agent,available]). However in many cases this is due to a network issue between the Zabbix server and the specific server being monitored. We'd like to be able to check the servers availability from two locations and only have the trigger fire if the check fails from both.

We're confused about the best solution to accomplish this. We've looked into creating a distributed setup with multiple nodes and also proxies but this would appear to just move the server from which the checks are being done.

Or we could setup a new server running zabbix-agent which has a item for every server in the network (net.tcp.port[<ip>,port]) and then reconfigure the trigger for each server to only fire if both items fail.

What would be the simplest solution to this problem? I'm sure this must be a common issue but I can't find much discussion about it.

Thanks

[Koral]

Hi,
we are sometimes doing the thing you mentioned as last - net.tcp.port[<ip>,port]. But only sometimes as we have only corporate network localized mostly in one city so we are using one of monitored HA servers to check connections between them.
But if you only have 80 servers than I would go for it, it is not much work to do
More work could be if you would set up a node and configure the routing (except if all monitored hosts use only public IPs)