Only poll data from agent

i’ve tried the search, but i’ve not come across an definitive answer.
I have a Ubuntu 22.04 server i want to monitor using librenms, latest version. I want to do so with the agent (check_mk agent). Agent installed, no problem. SNMPD is -not- enabled on the ubuntu server.
I cant get librenms to just use the agent for everythinng and not SNMPD. I don’t understand why, the module is enabled on the node.
My conclusion now is, after testing: i was only able to get the server monitored by librenms and read the agent when i enable the unix-agent module AND had snmpd up and running on the server.
Whithout snmpd answering on the server, i could not get anything out of the server.

Can you help understand why snmpd is needed when the agent is installed?
is snmpd required in all cases and do i need to see the agent as something always being complementary to the snmpd polling?

1 Like

anyone has some insights?

Hi @JayST

The information to be extracted from the server to be monitored is collected locally by SNMP, which delivers it to the agent, and it sends it to your remote NMS.

What do you gain from this?

You free the LibreNMS host from load since it will not have to perform snmpwalk against your Ubuntu, and will simply receive monitoring information periodically.

That’s how I understand it, and that’s why I consider it convenient.

Regards

oh wow is that really how that works with the agent? i would not have guessed that.
So that means that when the librennms host knows there’s an agent installed, it will not poll for snmp itself directly? It will instruct the agent to do so locally on the host and return the data? Just trying to understand better.

Regarding the agent: i see the linux agent is getting some updates, not that much, but some. But the windows agent, it’s really old and i’d therefor say not an option to use imho. am i missing something here? do i need to pull a newer agent somewhere else perhaps? or do we just poll windows servers with snmp directly without using the agent preferably?

An agent is independent software that collects data and sends it to the NMS, to a port on which it listens.
You do not need to be queried by the NMS. The advantage there is, the NMS does not have to execute a query process (which it would normally do with an SNMP query).
The query work is performed on the monitored host, which alleviates the CPU load of LibreNMS.

Personally, on Windows Server, I use SNMP today.

“and sends it to the NMS”
wait, does the agent push data to the librenms host? or do you mean it collects data on the monitored hosts and makes it available on a port (6556) for a librenms host or poller to collect that data by requesting it on that port (pull)?

Since Check_MK, SNMPd, and other agents can be used, despite being components of applications other than LibreNMS, they allow them to collect a lot of information locally on the hosts that host them, even more than what SNMP can reach. For example, referring to applications.
Periodically, LibreNMS consults them (polling) receiving said information, as it does on those hosts that are monitored with the use of the SNMP protocol.
The LibreNMS model is client-server. LibreNMS plays an active role, querying remote servers for monitoring information.
The advantage is that a single query to the remote host replaces multiple equivalent SNMP queries, and also allows adding information.
Let’s say that it presents advantages with the reduction of CPU costs and bandwidth.

1 Like

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.