Deploy Observability Locally

The observability stack provides access to metrics and logging for your workloads. This guide walks through the installation and usage of observability with Threeport.

Prerequisites

You'll need a local Threeport control plane for this guide. Follow the Install Threeport Locally guide to set that up.

You'll also want to deploy a workload to get metrics and logs for. Follow the Deploy Workload Locally guide to deploy a sample WordPress app.

Note: for this guide, you will need to have kubectl installed at this time.

Configs

First, create a work space on your local file system:

mkdir threeport-observability-test
cd threeport-observability-test

Download a sample workload config as follows:

curl -O https://raw.githubusercontent.com/threeport/threeport/main/samples/observability/observability-stack.yaml

You now have the config for an observability stack on your file system. This is what it looks like.

ObservabilityStack:
  Name: observability
  KubernetesRuntimeInstance:
    Name: eks-remote
  MetricsEnabled: true
  LoggingEnabled: true
  # LokiHelmValues: |
  # PromtailHelmValues: |
  # KubePrometheusStackHelmValues: |
  # KubePrometheusStackHelmValuesDocument: kube-prometheus-stack-values.yaml
  # GrafanaHelmValues: |
  #   grafana.ini:
  #     auth.anonymous:
  #       enabled: true
  #       org_role: Admin

The commented fields indicate how the observability stack can be customized, but that is outside the scope of this guide. We're going to deploy it with the default settings.

The only change you should make is to set the KubernetesRuntimeInstance.Name field. Get the name of the K8s runtime instance/s you have available with this command:

tptctl get kubernetes-runtime-instances

Create Observability Stack

This command will create all the observability stack components:

tptctl create observability-stack --config observability-stack.yaml

Give Threeport a few minutes to get all the stack components up and running.

You can use this kubectl command to determine when the observability dashboard is up and running.

kubectl get po -A -l app.kubernetes.io/name=grafana

When the output shows STATUS: Running like this, you can proceed.

NAMESPACE                                                  NAME                                                              READY   STATUS    RESTARTS   AGE
observability-observability-dashboard-grafana-dasjscmzhg   observability-observability-dashboard-grafana-release-6bf4ppg8g   3/3     Running   0          2m55s

Connect to the Dashboard

When running locally, you can connect to the observability dashboard as follows.

First get the namespace for the dashboard:

DASHBOARD_NAMESPACE=$(kubectl get po -A -l app.kubernetes.io/name=grafana -o=jsonpath='{.items[0].metadata.namespace}')

Then, create a port forward to allow you to connect to the dashboard locally:

kubectl port-forward pod/$(kubectl get po -A -l app.kubernetes.io/name=grafana -o=jsonpath='{.items[0].metadata.name}') 3000:3000 -n $DASHBOARD_NAMESPACE

Now visit http://localhost:3000 in your browser.

This should bring you to the Grafan login. Enter admin for the username and password for the password.

Metrics

On the left, click on the hamburger menu and navigate to Dashboards then Kubernetes / Compute Resources / Pod.

At the top of the dashboard, you can select the namespace and pod for which to view metrics.

Select the namespace with the wordpress- prefeix and one of the pods for that application. You can view CPU and memory usage, among other metrics, on this page for the workload.

Click "Home" at the top-left of the screen.

Logs

To get logs for the WordPress app, click the hamburger menu next to "Home" at the top-left of the screen. Then select Connections > Data sources from the menu.

On the Data sources page, click "Explore" on the loki data source.

On the Explore page, under "Label filters" select "pod" from the "Select label" menu and select the "getting-started-wordpress" pod from the "Select value" menu.

Now, click the blue "Run query" button at the top-right of the screen.

You'll now be able to view the logs from the WordPress application.

Remove Observability Stack

You can now hit ctrl-c to on the port forward to terminate it.

The observability stack can be uninstalled with the following command.

tptctl delete observability-stack --config observability-stack.yaml

Clean Up

Finally, we can remove the files we downloaded.

cd ../
rm -rf threeport-observability-test

Summary

In this guide we walked through how to install the observability stack and then use the dashboard to get basic metrics and log output from the sample Wordpress application.