Distributed Monitoring

[leandro0.o]

Hello!

I have a lot of servers worldwide.
Currently I have 1 Zabbix Server and Zabbix Agent installed on each servers.
What is happening are some false positive. A determinate host is online, but the network route to Zabbix Server is very bad, so it fails many time during agent.ping and it triggers unreachable host.
I know that I can to install Zabbix Proxy, but the same issue will occurs if connection between Zabbix Agent and Zabbix Proxy fail.

What I want to do?
I want to have N servers monitoring a host and it call the trigger only if fail in N servers.

Example:
ZA = Zabbix Agent; ZP = Zabbix Proxy; ZS = Zabbix Server.
ZP1 requests to ZA1 the key agent.ping
ZP2 requests to ZA1 the key agent.ping
ZP3 requests to ZA1 the key agent.ping

If all the agent.ping for the 3 ZP are 0, ZS call the trigger.

[valentins.grins]

Hello,

Usual architecture with proxy considers that you have proxy located in your remote site where network connection and latency to the hosts is good.

Regarding your idea, you can't monitor single host by multiple proxies.

I would recommend you to play with treshold for trigger alarms to find suitable solution for you. For instance you can do the check every minute and trigger fire alarm after 5 minutes only if all 5 pings received no data.
Also have a look for active checks, maybe it can improve your situation or maybe you can check trigger dependencies(https://www.zabbix.com/documentation...s/dependencies).

[leandro0.o]

valentins.grins, thank you for reply!

This was an idea that I had. A proxy for each continent.

It's sad to know that I can't monitor single host by multiple proxies it would be very useful!

Currently I used default settings of Template Zabbix Agent.
Just changed Agent Ping's interval to 30s. There is some servers that call the trigger "unreachable host" (5m without reply) but it's a false positive because server is online, just losing a lot of packets to Zabbix Server.