Cascading Telegrafs but retaining node-specific information post cascade

Quite new here to Telegraf and exploring it’s usage. I am more familiar with SNMP MIB world-view, so please bear with my noobish questions.

Let’s say that I have a bunch of nodes that expose their node specific metrics through Telegraf. Also, let’s say that I do not wish to expose the management IP of each of these nodes. I suppose, I could use the method described in this answer to create a cascade of Telegrafs through which all metrics pass (first being the node itself, second being a centralized Telegraf). Since this results in a fan-in of metrics from several nodes of cluster into this centralized Telegraf, just wondering as to whether the node-information is retained in the metrics ? The 2nd level centralized Telegraf should still be able to emit VM level metrics for that node itself (just to see if that centralized Telegraf VM itself is getting overloaded or not), right ?

It’s good question, and will probably need some experimentation.

The socket listener plug-in looks like it may be a good fit, and appears it would allow to pass through messaging received with minimal (possibly none) changes.

If on the other hand you want to know which telegraf endpoint collected some datapoint, then you might need to try something more custom in your telegraf config , maybe using the Defaults plug-in to insert a field , eg telegraf-agent and set it to the host name of the server running the agent.
I’d be curious how this turns out. Let us know if you have further questions on this.

1 Like

Thanks for the answer and potential direction to look at. Will try to experiment around with this.