My task is to install and configure Telegraf on ca.20 Windows servers in so-called “Secure Zone”., expose the metrics on each server’s http port, get them pulled by Telegraf running on machine in the “Border Zone”, and expose the combined stream on its http portfor collection by Prometheus running “somewhere out there”, behind the border.
Is that possible at all?
Reasonable, or there is a better solution?
Should I use inputs.http plugin on the “Central Telegraf”?
I see the hostname in each machine’s stream. Will Prometheus be able to properly attribute the metrics to specific source, basing on these hostnames?
I heard about some kind of auto-discovery. Would it be possible, instead of hardcoding all input URLs, just specify the subnet/mask, and let Telegraf discover all listening 9273 ports on it?
So far, this “Central Telegraf” is exposing only it’s own metrics, even though I’ve added few URLs to [[inputs.http]], checked that all URLs are reachable from Central, passed the syntax --test, and restarted the Windows service Still see only local metrics on TCP 9273 of the "Central Telegraf"
Thank you Andrei
