Overview

Grafana is an open-source platform for analytics and monitoring. It’s an ideal choice to use in combination with an InfluxDB metric drain. Grafan is useful in a number of ways:

  • It makes it easy to build beautiful graphs and set up alerts.

  • It works out of the box with InfluxDB.

  • It works very well in a containerized environment like Aptible.

Set up

Deploying with Terraform

The easiest and recommended way to set up Grafana on Aptible is using the Aptible Metrics Terraform Module. This provisions Aptible metric drains with pre-built Grafana dashboards and alerts for monitoring RAM and CPU usage for your Aptible apps and databases. This simplifies the setup of metric drains so you can start monitoring your Aptible resources immediately, all hosted within your Aptible account. If you would rather set it up from scratch, use this guide.

Deploying via the CLI

Step 1: Provision a PostgreSQL database

Grafana needs a Database to store sessions and Dashboard definitions. It works great with PostgreSQL, which you can deploy on Aptible.

Step 2: Configure the database

Once you have created the PostgreSQL Database, create a tunnel using the aptible db:tunnel command, then connect using psql and run the following commands to create a sessions database for use by Grafana:

CREATE DATABASE sessions;

Then, connect to the newly-created sessions database:

\c sessions;

And finally, create a table for Grafana to store sessions in:

CREATE TABLE session (
        key       CHAR(16) NOT NULL,
        data      BYTEA,
        expiry    INTEGER NOT NULL,
        PRIMARY KEY (key)
);

Step 3: Deploy the Grafana app

Grafana is available as a Docker image and can be configured using environment variables. As a result, you can use Direct Docker Image Deploy to easily deploy Grafana on Aptible.

Here is the minimal deployment configuration to get you started. In the example below, you’ll have to substitute a number of variables:

  • $ADMIN_PASSWORD: Generate a strong password for your Grafana admin user.

  • $SECRET_KEY: Generate a random string (40 characters will do).

  • $YOUR_DOMAIN: The domain name you intend to use to connect to Grafana (e.g. grafana.example.com).

  • $DB_USERNAME: The username for your PostgreSQL database. For a PostgreSQL database on Aptible, this will be aptible.

  • $DB_PASSWORD: The password for your PostgreSQL database.

  • $DB_HOST: The host for your PostgreSQL database.

  • $DB_PORT: The port for your PostgreSQL database.

aptible apps:create grafana

aptible deploy --app grafana --docker-image grafana/grafana \
        "GF_SECURITY_ADMIN_PASSWORD=$ADMIN_PASSWORD" \
        "GF_SECURITY_SECRET_KEY=$SECRET_KEY" \
        "GF_DEFAULT_INSTANCE_NAME=aptible" \
        "GF_SERVER_ROOT_URL=https://$YOUR_DOMAIN" \
        "GF_SESSION_PROVIDER=postgres" \
        "GF_SESSION_PROVIDER_CONFIG=user=$DB_USERNAME password=$DB_PASSWORD host=$DB_HOST port=$DB_PORT dbname=sessions sslmode=require" \
        "GF_LOG_MODE=console" \
        "GF_DATABASE_TYPE=postgres" \
        "GF_DATABASE_HOST=$DB_HOST:$DB_PORT" \
        "GF_DATABASE_NAME=db" \
        "GF_DATABASE_USER=$DB_USERNAME" \
        "GF_DATABASE_PASSWORD=$DB_PASSWORD" \
        "GF_DATABASE_SSL_MODE=require" \
        "FORCE_SSL=true"

📘 There are many more configuration options available in Grafana. Review Grafana’s configuration documentation for more information.

Step 4: Expose Grafana

Finally, follow the How do I expose my web app on the Internet? tutorial to expose your Grafana app over the internet. Make sure to use the same domain you configured Grafana with ($YOUR_DOMAIN in the example above)!

Using Grafana

Step 1: Log in

Once you’ve exposed Grafana, you can navigate to $YOUR_DOMAIN to access Grafana. Connect using the username admin and the password you configured above (ADMIN_PASSWORD).

Step 2: Connect to an InfluxDB Database

Once logged in to Grafana, you can connect Grafana to an InfluxDB database by creating a new data source. To do so, click the Grafana icon in the top left, then navigate to data sources and click “Add data source”.

The following assumes you have provisioned an InfluxDB database. You’ll need to interpolate the following values

  • $INFLUXDB_HOST: The hostname for your InfluxDB database. This is of the form db-$STACK-$ID.aptible.in.

  • $INFLUXDB_PORT: The port for your InfluxDB database.

  • $INFLUXDB_USERNAME: The username for your InfluxDB database. Typically aptible.

  • $INFLUXDB_PASSWORD: The password.

These parameters are represented by the connection URL for your InfluxDB database in the Aptible dashboard and CLI. For example, if your connection URL is https://foo:bar@db-qux-123.aptible.in:456, then the parameters are:

  • $INFLUXDB_HOST: db-qux-123.aptible.in

  • $INFLUXDB_PORT: 456

  • $INFLUXDB_USERNAME: foo

  • $INFLUXDB_PASSWORD: bar

Once you have those parameters in Grafana, use the following configuration for your data source:

  • Name: Any name of your choosing. This will be used to reference this data source in the Grafana web interface.

  • Type: InfluxDB

  • HTTP settings:

    • URL: https://$INFLUXDB_HOST:$INFLUXDB_PORT.

    • Access: proxy

  • HTTP Auth: Leave everything unchecked

  • Skip TLS Verification: Do not select

  • InfluxDB Details: - Database: If you provisioned this InfluxDB database on Aptible and/or are using it for an InfluxDB database metric drain, set this to db. Otherwise, use the database of your choice. - User: $INFLUXDB_USERNAME - Password: $INFLUXDB_PASSWORD

Finally, save your changes.

Step 3: Set up Queries

Here are a few suggested queries to get started with an InfluxDB metric drain. These queries are designed with Grafana in mind. To copy those queries into Grafana, use the raw text editor mode in Grafana.

📘 In the queries below, $__interval and $timeFilter will automatically be interpolated by Grafana. Leave those parameters as-is.

RSS Memory Utilization across all resources

SELECT MAX("memory_rss_mb") AS rss_mb
FROM "metrics"
WHERE $timeFilter
GROUP BY
        time($__interval),
        "app", "database", "service", "host"
        fill(null)

CPU Utilization for a single App

In the example below, replace ENVIRONMENT with the handle for your environment and HANDLE with the handle for your app

SELECT MEAN("milli_cpu_usage") / 1000 AS cpu
FROM "metrics"
WHERE
        environment = 'ENVIRONMENT' AND
        app = 'HANDLE' AND
        $timeFilter
GROUP BY
        time($__interval),
        "service", "host"
        fill(null)

Disk Utilization across all Databases

SELECT LAST(disk_usage_mb) / LAST(disk_limit_mb) AS utilization
FROM "metrics"
WHERE
        "database" <> '' AND
        $timeFilter
GROUP BY
        time($__interval),
        "database", "service", "host"
        fill(null)

Grafana documentation

Once you’ve added your first data source, you might also want to consider following Grafana’s getting started documentation to familiarize yourself with Grafana.

📘 If you get an error connecting, use the aptible logs commands to troubleshoot.

That said, an error logging in is very likely due to not properly creating the sessions database and the session table in it as indicated in Configuring the database.

Upgrading Grafana

To upgrade Grafana, deploy the desired version to your existing app containers:

aptible deploy --app grafana --docker-image grafana/grafana:VERSION

📘 Doing a big upgrade? If you need to downgrade, you can redeploy with a lower version. Alternatively, you can deploy a test Grafana app to ensure it works beforehand and deprovisioned the test app once complete.