This page will guide you through the process of configuring your Prometheus server to use your Metricfire account.
This guide assumes you already have a Prometheus setup roughly like this:
It's OK if you have more Prometheus instances, or if you've already configured read/write, federation, or are using Grafana with them - none of that will be affected.
After you set up Metricfire, here's what your Prometheus setup will look like:
In your prometheus.yml, add a new remote_write and remote_read sections like this:
Don't forget to fill in your Metricfire API key, which you can find on your Metricfire account page when you're logged in.
Restart your Prometheus server to make these changes take effect. Now is a good time to watch the logs from your Prometheus process to make sure it's still happy.
If you didn't configure the bearer token or it isn't correct for some reason, you will see errors like this in your Prometheus log:
level=warn ts=2018-11-01T00:00:00.00Z caller=queue_manager.go:531 component=remote queue=0:https://remote.metricfire.com/write msg="Error sending samples to remote storage" count=59 err="server returned HTTP status 403 FORBIDDEN: Forbidden"
Once we start receiving traffic from your Prometheus instance, we'll start processing and storing it. There's also a handy traffic indicator on your Metricfire account page that tells you how recently we received any data. If that says we received data recently, then it's working!
That should be it! Now, when you query your local Prometheus as before, it will send your query to Metricfire too. We'll check our index, and serve any data we have that match your query.
When data expires from your local Prometheus server, that data will no longer disappear from your dashboards – instead, it will be served from the Metricfire copy, and combined with the leading edge data from your local server.