Is Sensu a Good Fit For My Needs?


#1

Here’s the ideal scenario

single agent running inside the LAN collects API data from several local machines (which we can not install anything on, but do not require auth for http API access either. only address, port and endpoint)

this agent ships data to sensu core. more importantly are the alerts (telegram, slack or email ok) when the numbers returned by the API for any of the machines drops/rises from specified thresholds.

eventually we would integrate grafana to visualize this, but foremost we are looking to monitor the health of these other machines via the agent and send it to a web interface.

A) is this something you’d recommend using sensu for?

B) is the data flow
API endpoints>proxy clients>plugin>client>core (remember I cannot install anything on the machines i want to monitor)
and all the data stays inside the machine where core is?


#2

Hello Gaia,

Your use cases seems like it would be a good fit for Sensu Core. I do have a few questions though and then I could provide you a scenario on how you would solve that with Sensu Core.

  • What types of responses are you getting from your API? Strings? Integers? Return codes? By your description, it sounds like metrics and then you would want to alert based on some thresholds.
  • By web interface, do you mean a home grown one or Sensu’s Uchiwa?

The data would stay inside the machine where core is, but it is worth noting that Sensu only keeps the last 21 check results. You would need a long term storage solution for check results (ELK stack, Splunk) or for straight metrics (Graphite, InfluxDB, OpenTSDB), and then as you think you’d do, visualize with Grafana. Depending on where those live, they may be off the Sensu Core machine.

Regards,
Richard.