Lab: Monitor a Kubernetes cluster with the Datadog Operator

What you will learn in this TP

• Deploy the Operator and the agent
• Monitor an application with Autodiscovery
• Troubleshoot with agent status

Introduction #

What you will learn in this section

• Deploy the Operator and the agent
• Monitor an application with Autodiscovery
• Troubleshoot with agent status

Context. The team is moving its applications to Kubernetes. You must get the same visibility as on VMs, without manual configuration per pod.

Prerequisites: Datadog account (the free trial is enough) and its API key, Docker, kind, kubectl, helm  ·  Estimated time: 45 min

A cluster monitored declaratively.

Hands-on #

Guided steps, with their solution.

1. 1. Create the cluster and install the Operator

   Why? The Operator manages the agent from a single declarative resource.

   Create a kind cluster, install the Operator and the secret holding the API key.

2. 2. Deploy the agent

   Why? The DatadogAgent resource describes everything: site, key, features.

   Write and apply datadog-agent.yaml. How many agent pods do you get?

3. 3. Deploy an application

   Why? Autodiscovery annotations configure the check as soon as the pod starts.

   Deploy Redis with the redisdb check annotations and the unified tagging labels.

4. 4. Check on the agent side

   Why? agent status and configcheck tell whether the configuration was found and whether the check runs.

   Check that the Redis check is discovered and running.

5. 5. Check in Datadog

   Why? Unified tagging links metrics, logs and container.

   In Datadog, filter on service:cache: find the redis.* metrics, the container logs and the pod in the Kubernetes explorer.

Difficulty level : ●●●○○ (3/5)