Connecting two Zabbix servers

[mariob]

Hi to all. I am totally new in Zabbix and generally monitoring systems. Can you please help me.

Is it possible to connect two zabbix servers like in scheme in attachment? If it is, can you please point me where to find an implementation solution.

Thank you

Regards

Mario

[Colttt]

Hi,

it was possible in zabbix version 2.2 and earlier.. but in 2.4 and later it was dropped..

https://www.zabbix.com/documentation/2.2/manual/distributed_monitoring/nodes

https://www.zabbix.com/documentation/2.4/manual/introduction/whatsnew240#node-based_distributed_monitoring_removed

[Koffer]

Hi,

Basically you could do similar thing with an Agent

https://www.zabbix.com/documentation/3.2/manual/concepts/agent

[kowales]

Hi to all. I am totally new in Zabbix and generally monitoring systems. Can you please help me.

Is it possible to connect two zabbix servers like in scheme in attachment? If it is, can you please point me where to find an implementation solution.

Thank you

Regards

Mario

Hello Everyone!
I have similar task but in my case connection between Location 1 and Location 2 is not permament. My Location 2 is separated network with possibility to connect to the Intenet/VPN periodically (eg. scheduled).

This setup is for some industrial networks with local Zabbix servers that are maintained externally. We would like to get diagnostics from multiple sites, but with access to standard Zabbix eperience locally in sites.

What would be suggested solution for such setup?

Thanks for the help!
Kowales

[LenR]

Ideas only... having intermittent network connection gives you lots of snowflake (unique) concerns.

You could periodically send copies your remote zabbix DB's to a central location. You can access a zabbix DB from a frontend without a live zabbix server in the mix. Those db's could share a db server, the frontends could share a http stack with some config tweaks.

I think Grafana could then define each db/frontend as a data source.

You could use the API to get info out of each instance. For example, if you look at the pyzabbix api, you see calls like zapi.host.get. zapi is just an object, you can replace zapi with z1, repeat with z2-n and talk to multiple zabbix web frontends at the same time. (I do this to robo copy config items between a main and fall-back zabbix) This method works, but makes my head hurt actually writing the code.

[cesarsj]

I'm here to resurrect another post. In my case, I would like two servers to have monitoring redundancy in cases where the main server should stop for maintenance. My question is whether the agent can send data to two servers at the same time. Or would I have to run two instances of agent (I don't think it's the best option)?

[LenR]

Yes, you can have multiple entries in the agent Server and ServerActive parameters. The only issue I've really encountered is if you use zabbix_sender's -c option, it will only send to the first server listed in ServerActive. Any use of zabbix_sender will have to iterate over the list of target servers.

We do that, but the second Zabbix instance checks very few items, it's basically an up/down test, it doesn't watch usage or apps, but that's our design.

[cesarsj]

Interesting. Do you know of other side effects of deploying two zabbix servers to the same agent? I will discuss this with my boss.

[LenR]

No, we have had several cases where some hosts had dual monitoring, for migrations, customer groups had their own zabbix and the one above. "It just works" :-)

[1berto]

I use a 'two servers', independent locations, 5 + 5 proxies, 10.000 hosts and everything works flawlessly.
It's a breeze to make changes in the structure.