Auto discovery - linux servers

[bloodyskullz]

I did change the hostname recently but I restarted the service by typing

systemctl restart zabbix-agent2.service

in the agent conf, the passive and active are both set to the server IP it should be reporting to. If you want screenshots of something I can upload them.

[bloodyskullz]

So I just installed the agent on a windows server vm and it auto added to the hosts immediately.

If the linux VM had the agent installed and configured before the auto discovery rules on the server were setup, will it still auto discover the linux vm?

[kyus]

Yes, it would still discover your hosts.

Try to add your linux host manually without any interfaces and link the template "Linux by Zabbix agent active", see if it collect metrics... That way you can be sure that your configuration is setup correctly and there's no communication issues between agent and server.

Also make sure that the hostname you put in the frontend is the same as you defined in your agent config.

[bloodyskullz]

Yea that didn't do the trick



I am going to try another linux server and see if its just a one off.

[kyus]

If monitoring your VM with active items didn't work, then you have some issue with the communication between your agent and server... That's why autoregistration isn't working.

Try to check logs from agent and server.

[bloodyskullz]

The agent server is using agent 2 (which I believe is the newer one??) and i went to /var/log/zabbix and there is no file at all. If it's not there, what directory am I checking?

[kyus]

Well, that's the default path for the log file...

Your agent is running without any problems?
Try checking the journal for the service.

journalctl -xeu zabbix-agent2.service

Also, check the permissions for user zabbix in the log folder

[bloodyskullz]

I think there is an issue with the service but I can't tell:

journalctl -xeu zabbix-agent2.service

Feb 04 15:08:42 zabbix-agent zabbix_agent2[804]: 2025/02/04 15:08:42.919888 [NVIDIA] failed to kill plugin /usr/libexec/zabbix/zabbix-agent2-plugin-nvidia-gpu: Failed to kill plug>
Feb 04 15:08:42 zabbix-agent zabbix_agent2[804]: zabbix_agent2 [804]: ERROR: Cannot register plugins: failed to register metrics of plugin "NVIDIA": failed to start plugin: failed>
Feb 04 15:08:42 zabbix-agent systemd[1]: zabbix-agent2.service: Main process exited, code=exited, status=1/FAILURE
░░ Subject: Unit process exited
░░ Defined-By: systemd
░░ Support: https://www.debian.org/support
░░
░░ An ExecStart= process belonging to unit zabbix-agent2.service has exited.
░░
░░ The process' exit code is 'exited' and its exit status is 1.
Feb 04 15:08:42 zabbix-agent systemd[1]: zabbix-agent2.service: Failed with result 'exit-code'.
░░ Subject: Unit failed
░░ Defined-By: systemd
░░ Support: https://www.debian.org/support
░░
░░ The unit zabbix-agent2.service has entered the 'failed' state with result 'exit-code'.

However when I run systemctl status zabbix-agent2.service, I get this output:

● zabbix-agent2.service - Zabbix Agent 2
     Loaded: loaded (/lib/systemd/system/zabbix-agent2.service; enabled; preset: enabled)
     Active: active (running) since Tue 2025-02-04 15:08:39 EST; 191ms ago
   Main PID: 804 (zabbix_agent2)
      Tasks: 6 (limit: 2261)
     Memory: 5.7M
        CPU: 76ms
     CGroup: /system.slice/zabbix-agent2.service
             └─804 /usr/sbin/zabbix_agent2 -c /etc/zabbix/zabbix_agent2.conf
​

[kyus]

Your service was up for 191ms

Isn't it in a restart loop?

Since the error you're getting is from this nvidia plugin, just disable it, there's been more people relating problems because of that.

Edit this file: /etc/zabbix/zabbix_agent2.d/plugins.d/nvidia.conf
Then comment out the following line:

# Plugins.NVIDIA.System.Path=/usr/libexec/zabbix/zabbix-agent2-plugin-nvidia-gpu
Save the file and restart the Zabbix Agent 2 service. The error should no longer appear.

Here's the discussion about this topic:

https://www.zabbix.com/forum/zabbix-help/496548-zabbix-agent2-7-2-1-fails-to-start-with-nvidia-gpu-plugin-on-linux

[bloodyskullz]

Yea that did the trick, plugin was the issue. I didn't think my nvidia GPU as being passed through to the vm.

So the conf file must be edited every time and the hostname + the passive and active check parameters need to be set for the agent to talk to the server?

If someone has hundreds of VM's, how does it get deployed?

[cyber]

For hundreds of servers you should have tools for deploying.. Puppet, Ansible, Terraform etc etc..

[bloodyskullz]

Thank you to everyone who assisted in supporting this. I will create another topic regarding what I am looking to do.