NPM "Agent" on Linux: How to add additional "custom metrics" or monitoring ... plugins or use SAM templates?

tigger2

I'm not seeing anything about this, so I just wanted to check my sanity.

Background:

We have installed an "Agent" on a Linux server.  Its collecting a bunch of "standard" metrics (like CPU/memory/etc) that we get on a windows server when we connect via WMI and we list the "resources" to monitor.

It appears the agents are configured? for metrics collection based on "plug-ins". I can "view the plugins" from the 'manage agents" page in the NPM web UI. 

Our agent is configured as 'passive", meaning our Orion poller reaches out using a "shared secret" to get data from the agent (my assumption is that, in this mode, the agent does nothing until the Poller reaches out to it to initiate data collection)

Questions:

If we wanted to collect "more" data than is being currently collected *using ONLY the agent*, it appears we cannot (or do not) download any extra "plugins". It appears the plugins" are "whatever comes with the product" and you do not normally develop plugins on your own or download them from somewhere (like you can with SAM templates)? 

It appears that we can use SAM templates, but I thought these 'connect to the server via ssh and run scripts every poll" on Linux vs "pushes the script down to the agent and it stays there in the agent, ready for execution" (.  But I see inside SAM templates there is an option under 'advanced" for "Preferred Polling Method: Agent"... so if this is chosen do templates get pushed to agents 1 time and stay there, ready for execution?

weston.hood

So first, your assumption is correct in passive mode. The agent sits there waiting for a connection from the assigned polling engine to get polling jobs.

As far as Plug-ins go, you should never need to manage these. That's more for tech support when there's an issue. But plug-ins automatically get deployed based on the need for them. For example, it's probably not going to receive a SAM application plugin until you assign a template to the Linux machine. Then it will auto-deploy the appropriate plugins. Or if you enabled application dependency mapping, for example, it would/should deploy additional plugins to collect the specific information for external connection mapping. All needed plugins ship with the software that you install, so again, you don't really need to manage the plugins or their deployment, it is simply pushed out as needed.

The default "preferred polling method" in SAM templates is agent, you are correct. But you can optionally flip it to be non-agent polled. Which would effectively not use the agent but instead do what you stated - it would perform an SSH session and run the template remotely (if script) and/or do it via SNMP if something like process monitor, etc. But most likely, you're going to want to have the agent run the script. Sometimes there are advantages to having the agent do the job, especially on Windows because the agent runs as a Local System account, and that can sometimes bypass GPO or permissions needed to run something. Also bypasses the need to set up WinRM. The agent has a local JobEngine, so if you assign a template, it would (in your case) get pushed out via the polling engine to the agent and run that job until the assigned engine orders otherwise.