Monitoring servers over internet

[ralfaro]

Hello everyone, new to the forum and new to Zabbix

I'm evaluating Zabbix to use on our environment, we where using a SaaS solution to monitor our servers, but a recent change on terms on their side, and new business needs made it difficult to keep using their product, so we decided to look for other solution and move the monitoring solution on-premise.

Our requirements are these:

- Agent based monitoring that can work over the internet without the need of a VPN, opening ports or any other changes on the monitored server side, basically we need the agent to contact the monitoring server over the internet and send the collected metrics.

- SNMP suport is not mandatory.

- Ability to send alerts over email.

- Ability to collect custom metrics from 3rd parties, like AWS CloudWatch.

- We should be able to pull data from the database (in this case Zabbix's database) and use it for custom applications, like reports, feed websites, 3rd party apps and so on.

I need your help to know if we can achive that with Zabbix.

I already have a lab running Zabbix on RHEL 7, so far I have been able to install and configure the server, as well as run the agent on several servers on the same network where the Zabbix server is, but I've not been able to make the agent connect to the server over the Internet, what should I check?

Thanks all for your help.

Cheers.

[elkato]

Hi ralfaro,

Regards the network, for what I know, as default Zabbix works with passive checks, it means Zabbix server connect to the zabbix-agents checks and return with data.
Over the network, this is a connectiion to a TCP port defined. By default TCP 10050, but you can change it. So If that port is opened NATed/routed to the server you wouldn't have any problem.

The same with SNMP (use UDP 161 by default).

I will secure the connections, allowing only the authorized IP of the zabbix server.

[jamesNJ]

On your first bullet item there are a few things you could do. See this recent thread:

https://www.zabbix.com/forum/showthread.php?t=25934

You could possibly use either active checks or maybe a proxy at the remote site.

I think it might be simpler if you could arrange for the monitored host to open up a port for inbound communication. See my response on page 2 for the thread above. I was able to very easily use stunnel to create a secure connection to my internet monitored hosts. The firewall at my cloud host allows to set inbound port to a specific IP address, further adding to security.

If both endpoints are linux/unix, you could also arrange for the monitored host to initiate a persistent SSH tunnel to your zabbix host so that zabbix can use the usual agent connections.

[ralfaro]

Thanks for the answers guys, I'll try with your suggestions. Cheers.