Configure Prometheus remote write to send metrics to Grafana Cloud

In this milestone, you add a remote_write block to your Prometheus configuration file, usually prometheus.yml. Grafana Cloud generates the block for you on the Hosted Prometheus metrics setup page. It contains the /api/prom/push URL of your Grafana Cloud Metrics instance, your instance ID as the username, and an access policy token as the password.

You don’t need to leave your Grafana Cloud stack for any of this. The same values are also available from the Prometheus card in the Cloud Portal.

First, open the setup page for hosted Prometheus metrics.

1. In the navigation menu, click Connections.

2. Click Add new connection.

3. In the search field, enter Hosted Prometheus to filter the catalog.

4. Click the Hosted Prometheus metrics tile.

The setup page is open. It walks you through choosing a method, creating a token, and copying the configuration.

Next, choose the remote write method and create a token. Creating a token needs the Admin role in your stack.

1. Under Choose a method for forwarding metrics, select From my local Prometheus server.

   The page preselects Via Grafana Alloy, which installs a separate collector. This path uses the remote write built into Prometheus instead.

2. Under Choose your use case for forwarding metrics, select Send metrics from a single Prometheus instance.

   The other cards cover multiple or high-availability Prometheus servers. The page adjusts its instructions if you pick one of those.

3. Under Select how to send the metrics from Prometheus to Grafana Cloud, keep Directly selected.

4. Under Update Prometheus configuration, keep Create a new token selected and enter a Token name you’ll recognize, for example prometheus-remote-write.

5. Click Create token.

   Grafana Cloud creates an access policy token and inserts it as the password in the remote_write configuration that appears below. The token is only shown once, so keep this page open until you’ve copied the configuration.

6. Below the configuration, click Copy to clipboard.

   The block contains the URL of your Grafana Cloud Metrics instance, your instance ID as the username, and the token as the password.

You have a remote_write block with your stack’s endpoint and credentials.

Now add it to Prometheus. These steps happen on the machine that runs Prometheus, so this path can’t detect them. When Prometheus restarts cleanly, mark the section as complete to continue.

1. Log in to the machine on which Prometheus is installed, and open a terminal.

2. If your user can’t write to the Prometheus configuration file, switch to one that can.

   For example:

   Copy

   sudo su -

3. Open prometheus.yml in a text editor. On many Linux installations the file is in /etc/prometheus.

   For example, with Vim, run:

   Copy

   vim /etc/prometheus/prometheus.yml

4. Scroll to the end of the file and paste the remote_write block you copied.

   In Vim, press i to start editing before you paste. The result looks like this, with your own values in place of the placeholders:

   YAML Copy

   remote_write:
     - url: <Your Metrics instance remote_write endpoint>
       basic_auth:
         username: <Your Metrics instance ID>
         password: <Your Cloud Access Policy token>

   If the file already has a remote_write section, add the new entry to that list instead of adding a second remote_write key.

5. Save the file and exit the editor.

   In Vim, press Esc, then type :wq and press Enter.

6. Restart Prometheus so it reads the new configuration.

   For example, if Prometheus runs as a systemd service:

   Copy

   systemctl restart prometheus

   After the restart, Prometheus starts pushing the samples it scrapes to Grafana Cloud.

7. Check that Prometheus restarted without errors:

   Copy

   systemctl status prometheus

   If the service failed to start, a common cause is a YAML indentation error in the block you pasted. Check the service logs with journalctl -u prometheus or in /var/log/syslog.

Prometheus now has a second destination for the samples it scrapes. Your scrape configuration, recording rules, and alerts don’t change.

In the next milestone, you verify that your metrics are arriving in Grafana Cloud.

More to explore (optional)

• Send data from multiple Prometheus instances
• Send data from multiple high-availability Prometheus instances
• Send Prometheus native histograms