Grafana Persistence

Grafana Persistence

Here we will talk about how to connect our instances of Grafana to a database on Gcloud to ensure that data such as logins, permissions, and dashboards persist if the pod stops or is restarted for whatever reason.

Setup

1. First, we need a database that Grafana will link to. To do this go to the Google Cloud console and in the menu go into the SQL section.

2. Look for an instance named aimmo-monitoring, if this does not exist for whatever reason, one will need to be created by following the steps after clicking the create instance button (you can create either a MySQL or PostgreSQL instance, for simplicity we recommend choosing PostgreSQL).

3. Now that you have an instance, go into it, then find the databases tab. You need 1 database per instance of Grafana (so one per cluster), the names of these databases are used later by the Grafana deployment so they are important. Recommended names are: aimmo_monitoring_dev, aimmo_monitoring_staging, aimmo_monitoring_default.

4. You should now have a database for the cluster you're working with. Next, you need a user, there should already be a user called grafana that you can use, if not click create user and go through the steps there. You need to make sure the user will accesss to the instance you created if you had to create a new one.

5. next, you should see 2 secrets on the cluster, one contains service account info, and the other has login details for grafana. these are used in the grafana-deployment.yaml. If you need to recreate these secrect for whatever reason make sure the correct data is passed into this yaml. Here is an example deployment which makes a good reference for this step.