prom_ex VS agent

This post is about another cool Fly.io feature --- built-in Prometheus metrics --- and how easy it is to take advantage of them in an Elixir application. I wrote and maintain an Elixir library, PromEx, that makes it a snap to export all sorts of metrics from your Elixir applications and get them on dashboards in Grafana. Let's explore some of the concepts surrounding Prometheus and see how we can leverage the Fly.io monitoring tools in an Elixir application to get slick looking dashboards like this one:

And has an agent that can be run on the host to monitor it: https://github.com/grafana/agent

For example, there are things like https://github.com/grafana/agent

Hey, I would like to lint my agent config in vscode. Did someone know where I can find the agent config JSON schema ? I've searched on https://github.com/grafana/agent and https://grafana.com/docs/agent/latest/ but cant find anything like this ...

Grafana Agent can stream data back to you.

At the time, Jaeger did not support OTEL format, which meant we needed our telemetry collector to export spans in Jaeger format. We were able to push this small fix to add a Jaeger exporter on the Agent.

Furthermore, would recommend Grafana Agent OR Prometheus Agent in this case since you probably don't need the Prometheus UI in each Cluster as well as the Alerting stuff that is inside Prometheus. (Mimir will do the ruling stuff for you). Grafana Agent also has an Operator mode if you want to use ServiceMonitor and PodMonitor CustomResources.

There were a bunch of alternatives to solve this before, like grafana agent [0], vm agent [1] or opentelemetry collector [2].

[0] - https://github.com/grafana/agent

[1] - https://github.com/VictoriaMetrics/VictoriaMetrics/tree/mast...

[2] - https://github.com/open-telemetry/opentelemetry-collector

This might be a reasonable use case for the Grafana Agent. It's basically a Prometheus instance, but without the local storage. Just forwards the data.