Skip to content

OpenTelemetry observability

Keptn makes any Kubernetes deployment observable. In other words, it creates a distributed, end-to-end trace of what Kubernetes does in the context of a Deployment. To do this, Keptn introduces the concept of an application, which is an abstraction that connects multiple Workloads that logically belong together, even if they use different deployment strategies.

This means that:

  • You can readily see why a deployment takes so long or why it fails, even when using multiple deployment strategies.
  • Keptn can capture DORA metrics and expose them as OpenTelemetry metrics

The observability data is an amalgamation of the following:

  • DORA metrics are collected out of the box when Keptn is enabled
  • OpenTelemetry runs traces that show everything that happens in the Kubernetes cluster
  • Custom Keptn metrics that you can use to monitor information from all the data providers configured in your cluster

All this information can be displayed with dashboard tools such as Grafana.

For an introduction to using OpenTelemetry with Keptn metrics, see the Keptn Observability getting started guide.

Requirements for OpenTelemetry

To access OpenTelemetry metrics with Keptn, you must have the following on your cluster:

  • An OpenTelemetry collector. See OpenTelemetry Collector for more information.
  • A Prometheus Operator. See Prometheus Operator Setup.

  • To install Prometheus into the monitoring namespace using the example configuration included with Keptn, use the following command sequence. You can modify these commands to define a different configuration:

    Note You must clone the lifecycle-toolkit repository and cd into the correct directory (examples/support/observability) before running the following commands.

    kubectl create namespace monitoring
    kubectl apply --server-side -f config/prometheus/setup/
    kubectl apply -f config/prometheus/
    
  • If you want a dashboard for reviewing metrics and traces:

    • Install Grafana or the visualization tool of your choice, following the instructions in Grafana Setup.
    • Install Jaeger or a similar tool for traces following the instructions in Jaeger Setup.

    • Follow the instructions in the Grafana README file to configure the Grafana dashboard(s) for Keptn.

Metrics can also be retrieved without a dashboard. See Accessing Metrics via the Kubernetes Custom Metrics API

Integrate OpenTelemetry into Keptn

To integrate OpenTelemetry into Keptn:

  • Apply basic annotations for the Deployment resource(s) to integrate Keptn into your Kubernetes cluster.
  • To expose OpenTelemetry metrics, define a KeptnConfig resource that has the spec.OTelCollectorUrl field populated with the URL of the OpenTelemetry collector.

The otel-collector.yaml is the OpenTelemetry manifest file for the PodtatoHead example, located in the config directory. To deploy and configure the OpenTelemetry collector using this manifest, the command is:

kubectl apply -f config/otel-collector.yaml \
    -n keptn-system

Use the following command to confirm that the pod for the otel-collector deployment is up and running:

$ kubectl get pods -lapp=opentelemetry \
    -n keptn-system

NAME                              READY   STATUS    RESTARTS      AGE
otel-collector-6fc4cc84d6-7hnvp   1/1     Running   0             92m

If you want to extend the OTel Collector configuration to send your telemetry data to other Observability platform, you can edit the Collector ConfigMap with the following command:

kubectl edit configmap otel-collector-conf \
    -n keptn-system

When the otel-collector pod is up and running, restart the keptn-scheduler (if installed) and lifecycle-operator so they can pick up the new configuration:

kubectl rollout restart deployment \
    -n keptn-system keptn-scheduler lifecycle-operator

Keptn begins to collect OpenTelemetry metrics as soon as the Deployment resource has the basic annotations to integrate Keptn in the cluster.

Access Keptn metrics as OpenTelemetry metrics

Keptn metrics can be exposed as OpenTelemetry (OTel) metrics via port 9999 of the Keptn metrics-operator.

To access the metrics, use the following command:

kubectl port-forward deployment/metrics-operator 9999 -n keptn-system

You can access the metrics from your browser at: http://localhost:9999

Advanced tracing configurations in Keptn: Linking traces

In Keptn you can connect multiple traces, for instance to connect deployments of the same application through different stages. To create connections between the traces of versions of your application, you can enrich the KeptnAppContext resource with OpenTelemetry span links. You can retrieve the span link from the JSON representation of the trace in Jaeger, where it has the following structure:

00-<trace-id>-<span-id>-01

Use this value to populate the spanLinks field of your KeptnAppContext resource to connect traces of different versions of the application.

apiVersion: lifecycle.keptn.sh/v1beta1
kind: KeptnAppContext
metadata:
  name: keptndemoapp
  namespace: keptndemo
spec:
  metadata:
    commit-id: "1234"
    author: "myUser"
  spanLinks:
    - "00-c088f5c586bab8649159ccc39a9862f7-f862289833f1fba3-01"

Note The span link used above is just an example. You need to replace it with the traceID and spanID that you retrieve from the KeptnAppVersion resource you want to link.

To store this new information in the traces, you need to increment the version of your application and apply theKeptnAppContext. Keptn will re-deploy your application and Jaeger should show a link to the previous trace in the references section.

linked trace

Comments