OpenTelemetry Metrics#

In this lab module we'll utilize the OpenTelemetry Collector deployed as a DaemonSet (Node Agent) to collect Node (kubelet) metrics from a Kubernetes cluster and ship them to Dynatrace. Additionally, we'll utilize a second OpenTelemetry Collector deployed as a Deployment (Gateway) to collect Cluster (Kubernetes API) metrics from the Kubernetes cluster and ship them to Dynatrace.

Lab tasks:

Deploy OpenTelemetry Collector as a DaemonSet
Configure OpenTelemetry Collector service pipeline for metric enrichment
Deploy OpenTelemetry Collector as a Deployment
Configure OpenTelemetry Collector service pipeline for metric enrichment
Query and visualize metrics in Dynatrace using DQL

Learn More

Prerequisites#

Import Notebook into Dynatrace

Notebook

Define workshop user variables

In your Github Codespaces Terminal set the environment variables:

Sprint Environment

Are you using a Sprint environment for your Dynatrace tenant? If so, then use export DT_ENDPOINT=https://{your-environment-id}.sprint.dynatracelabs.com/api/v2/otlp instead of the live version below.

export DT_ENDPOINT=https://{your-environment-id}.live.dynatrace.com/api/v2/otlp
export DT_API_TOKEN={your-api-token}
export NAME=<INITIALS>-k8s-otel-o11y

Move into the metrics module directory

Command:

cd $BASE_DIR/lab-modules/dt-k8s-otel-o11y-metrics

Collector for Node Metrics#

Kubernetes Node Metrics

Each Kubernetes Node runs a kubelet that includes an API server. The kubeletstats Receiver connects to that kubelet via the API server to collect metrics about the node and the workloads running on the node.

Deploy OpenTelemetry Collector#

Contrib Distro - Daemonset (Node Agent)

The kubeletstats receiver is only available on the Contrib Distro of the OpenTelemetry Collector. Therefore we must deploy a new Collector using the contrib image.

---
apiVersion: opentelemetry.io/v1alpha1
kind: OpenTelemetryCollector
metadata:
  name: dynatrace-metrics-node
  namespace: dynatrace
spec:
  envFrom:
  - secretRef:
      name: dynatrace-otelcol-dt-api-credentials
  env:
    - name: K8S_NODE_NAME
      valueFrom:
        fieldRef:
          fieldPath: spec.nodeName
  mode: "daemonset"
  image: "otel/opentelemetry-collector-contrib:0.103.0"

Command:

kubectl apply -f opentelemetry/collector/metrics/otel-collector-metrics-node-crd-01.yaml

Sample output:

opentelemetrycollector.opentelemetry.io/dynatrace-metrics-node created

Validate running pod(s)

Command:

kubectl get pods -n dynatrace

Sample output:

NAME	READY	STATUS	RESTARTS	AGE
dynatrace-metrics-node-collector-2kzlp	1/1	Running	0	1m

Configure Kubernetes RBAC#

Create clusterrole with read access to Kubernetes objects

Since the receiver uses the Kubernetes API, it needs the correct permission to work correctly. For most use cases, you should give the service account running the Collector the following permissions via a ClusterRole.

---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: otel-collector-k8s-clusterrole-metrics
rules:
  - apiGroups: ['']
    resources: ['events', 'namespaces', 'namespaces/status', 'nodes', 'nodes/spec', 'nodes/stats', 'nodes/proxy', 'pods', 'pods/status', 'replicationcontrollers', 'replicationcontrollers/status', 'resourcequotas', 'services']
    verbs: ['get', 'list', 'watch']
  - apiGroups: ['apps']
    resources: ['daemonsets', 'deployments', 'replicasets', 'statefulsets']
    verbs: ['get', 'list', 'watch']
  - apiGroups: ['extensions']
    resources: ['daemonsets', 'deployments', 'replicasets']
    verbs: ['get', 'list', 'watch']
  - apiGroups: ['batch']
    resources: ['jobs', 'cronjobs']
    verbs: ['get', 'list', 'watch']
  - apiGroups: ['autoscaling']
    resources: ['horizontalpodautoscalers']
    verbs: ['get', 'list', 'watch']

Command:

kubectl apply -f opentelemetry/rbac/otel-collector-k8s-clusterrole-metrics.yaml

Sample output:

clusterrole.rbac.authorization.k8s.io/otel-collector-k8s-clusterrole-metrics created

Create clusterrolebinding for OpenTelemetry Collector service account

---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: otel-collector-k8s-clusterrole-metrics-crb
subjects:
- kind: ServiceAccount
  name: dynatrace-metrics-node-collector
  namespace: dynatrace
roleRef:
  kind: ClusterRole
  name: otel-collector-k8s-clusterrole-metrics
  apiGroup: rbac.authorization.k8s.io

Command:

kubectl apply -f opentelemetry/rbac/otel-collector-k8s-clusterrole-metrics-crb.yaml

Sample output:

clusterrolebinding.rbac.authorization.k8s.io/otel-collector-k8s-clusterrole-metrics-crb created

`kubeletstats` receiver#

OpenTelemetry Documentation

By default, metrics will be collected for pods and nodes, but you can configure the receiver to collect container and volume metrics as well. The receiver also allows configuring how often the metrics are collected:

config: |
    receivers:
      kubeletstats:
        collection_interval: 30s
        auth_type: 'serviceAccount'
        endpoint: '${env:K8S_NODE_NAME}:10250'
        insecure_skip_verify: true
        metric_groups:
          - node
          - pod
          - container

Default Metrics: OpenTelemetry Documentation

note: for this lab, the Kind cluster does not have cluster metadata to collect. These values will be spoofed for the purposes of this lab.

resource/kind:
  attributes:
  - key: k8s.cluster.name
    value: dt-k8s-o11y-kind
    action: insert

Query Node metrics in Dynatrace

DQL:

timeseries node_cpu = avg(k8s.node.cpu.utilization), by: {k8s.cluster.name, k8s.node.name}

Result:

dql_kubeletstats_node_cpu

k8sattributes Processor#

The Kubernetes Attributes Processor automatically discovers Kubernetes pods, extracts their metadata, and adds the extracted metadata to spans, metrics, and logs as resource attributes.

The Kubernetes Attributes Processor is one of the most important components for a collector running in Kubernetes. Any collector receiving application data should use it. Because it adds Kubernetes context to your telemetry, the Kubernetes Attributes Processor lets you correlate your application’s traces, metrics, and logs signals with your Kubernetes telemetry, such as pod metrics and traces.

Add k8sattributes processor

OpenTelemetry Documentation

The k8sattributes processor will query metadata from the cluster about the k8s objects. The Collector will then marry this metadata to the telemetry.

k8sattributes:
  auth_type: "serviceAccount"
  passthrough: false
  filter:
    node_from_env_var: KUBE_NODE_NAME
  extract:
    metadata:
      - k8s.namespace.name
      - k8s.deployment.name
      - k8s.daemonset.name
      - k8s.job.name
      - k8s.cronjob.name
      - k8s.replicaset.name
      - k8s.statefulset.name
      - k8s.pod.name
      - k8s.pod.uid
      - k8s.node.name
      - k8s.container.name
      - container.id
      - container.image.name
      - container.image.tag
    labels:
      - tag_name: app.label.component
        key: app.kubernetes.io/component
        from: pod
    pod_association:
      - sources:
          - from: resource_attribute
            name: k8s.pod.uid
      - sources:
          - from: resource_attribute
            name: k8s.pod.name
      - sources:
          - from: resource_attribute
            name: k8s.pod.ip
      - sources:
          - from: connection

Command:

kubectl apply -f opentelemetry/collector/metrics/otel-collector-metrics-node-crd-02.yaml

Sample output:

opentelemetrycollector.opentelemetry.io/dynatrace-metrics-node configured

Validate running pod(s)

Command:

kubectl get pods -n dynatrace

Sample output:

NAME	READY	STATUS	RESTARTS	AGE
dynatrace-metrics-node-collector-drk1p	1/1	Running	0	1m

Query Pod metrics in Dynatrace

DQL:

timeseries pod_cpu = avg(k8s.pod.cpu.utilization), by: { k8s.pod.name, k8s.node.name, k8s.namespace.name, k8s.deployment.name, k8s.cluster.name, k8s.pod.uid }
| filter k8s.namespace.name == "astronomy-shop" and k8s.deployment.name == "astronomy-shop-productcatalogservice"

Result:

dql_kubeletstats_pod_cpu

Collector for Cluster Metrics#

The Kubernetes Cluster Receiver collects metrics and entity events about the cluster as a whole using the Kubernetes API server. Use this receiver to answer questions about pod phases, node conditions, and other cluster-wide questions.

Deploy OpenTelemetry Collector#

Contrib Distro - Deployment (Gateway)

OpenTelemetry Documentation

The k8s_cluster receiver is only available on the Contrib Distro of the OpenTelemetry Collector. Therefore we must deploy a new Collector using the contrib image.

Since the receiver gathers telemetry for the cluster as a whole, only one instance of the receiver is needed across the cluster in order to collect all the data. The Collector will be deployed as a Deployment (Gateway).

---
apiVersion: opentelemetry.io/v1alpha1
kind: OpenTelemetryCollector
metadata:
  name: dynatrace-metrics-cluster
  namespace: dynatrace
spec:
  envFrom:
  - secretRef:
      name: dynatrace-otelcol-dt-api-credentials
  mode: "deployment"
  image: "otel/opentelemetry-collector-contrib:0.103.0"

Command:

kubectl apply -f opentelemetry/collector/metrics/otel-collector-metrics-cluster-crd-01.yaml

Sample output:

opentelemetrycollector.opentelemetry.io/dynatrace-metrics-cluster created

Validate running pod(s)

Command:

kubectl get pods -n dynatrace

Sample output:

NAME	READY	STATUS	RESTARTS	AGE
dynatrace-metrics-cluster-collector-7bd8dc4995-6sgs2	1/1	Running	0	1m

`k8s_cluster` receiver#

OpenTelemetry Documentation

config: |
    receivers:
      k8s_cluster:
        collection_interval: 60s
        node_conditions_to_report: [ "Ready", "MemoryPressure", "DiskPressure" ]
        allocatable_types_to_report: [ "cpu","memory" ]
        metadata_collection_interval: 5m

Default Metrics: OpenTelemetry Documentation

Query Deployment metrics in Dynatrace

DQL:

timeseries pods_avail = min(k8s.deployment.available), by: {k8s.cluster.name, k8s.deployment.name}, filter: {k8s.namespace.name == "astronomy-shop"}

Result:

dql_k8scluster_pod_avail

Export Application Metrics#

The astronomy-shop demo application has the OpenTelemetry agents and SDKs already instrumented. These agents and SDKs are generating metrics (traces and logs too) that are being exported to a Collector running within the astronomy-shop namespace bundled into the application deployment. We want these metrics to be shipped to Dynatrace as well.

`otlp` receiver#

OpenTelemetry Documentation

Adding the otlp receiver allows us to receive telemetry from otel exporters, such as agents and other collectors.

config: |
    receivers:
      otlp:
        protocols:
          grpc:
            endpoint: 0.0.0.0:4317
          http:
            endpoint: 0.0.0.0:4318
    service:
      pipelines:
        metrics:
          receivers: [otlp]
          processors: [batch]
          exporters: [otlphttp/dynatrace]

Export OpenTelemetry data from astronomy-shop to OpenTelemetry Collector - Contrib Distro

Customize astronomy-shop helm values

OpenTelemetry data created by agents and SDKs should include service.name and service.namespace attributes. We will make the service.namespace unique to our deployment using our NAME environment variable declared earlier, using a sed command on the Helm chart's values.yaml file.

default:
  # List of environment variables applied to all components
  env:
    - name: OTEL_SERVICE_NAME
      valueFrom:
        fieldRef:
          apiVersion: v1
          fieldPath: "metadata.labels['app.kubernetes.io/component']"
    - name: OTEL_COLLECTOR_NAME
      value: '{{ include "otel-demo.name" . }}-otelcol'
    - name: OTEL_EXPORTER_OTLP_METRICS_TEMPORALITY_PREFERENCE
      value: cumulative
    - name: OTEL_RESOURCE_ATTRIBUTES
      value: 'service.name=$(OTEL_SERVICE_NAME),service.namespace=NAME_TO_REPLACE,service.version={{ .Chart.AppVersion }}'

service.namespace=NAME_TO_REPLACE\ service.namespace=INITIALS-k8s-otel-o11y

Command:

sed -i "s,NAME_TO_REPLACE,$NAME," astronomy-shop/collector-values.yaml

Update astronomy-shop OpenTelemetry Collector export endpoint via helm

Our collector-values.yaml contains new configurations for the application so that the astronomy-shop Collector includes exporters that ship to the Collectors deployed in the dynatrace namespace.

exporters:
  # Dynatrace OTel Collectors
  otlphttp/dttraces:
    endpoint: http://dynatrace-traces-collector.dynatrace.svc.cluster.local:4318
  otlphttp/dtlogs:
    endpoint: http://dynatrace-logs-collector.dynatrace.svc.cluster.local:4318
  otlphttp/dtmetrics:
    endpoint: http://dynatrace-metrics-cluster-collector.dynatrace.svc.cluster.local:4318

Command:

helm upgrade astronomy-shop open-telemetry/opentelemetry-demo --values astronomy-shop/collector-values.yaml --namespace astronomy-shop --version "0.31.0"

Sample output:

NAME: astronomy-shop\ LAST DEPLOYED: Thu Jun 27 20:58:38 2024\ NAMESPACE: astronomy-shop\ STATUS: deployed\ REVISION: 2

Query astronomy-shop metrics in Dynatrace

DQL:

timeseries jvm_mem_used = avg(jvm.memory.used), by: {service.name, k8s.cluster.name}, filter: {k8s.namespace.name == "astronomy-shop"}

Result:

dql_sdk_jvm_mem

DQL:

timeseries avg(kafka.consumer.request_rate), by: {service.name, k8s.cluster.name}, filter: {k8s.namespace.name == "astronomy-shop"}

Result:

dql_sdk_kafka_request_rate

Browse available metrics in Dynatrace

You can browse all available metrics from OpenTelemetry sources in the Metrics Browser. Filter on Dimension:otel.scope.name to find relevant metrics.

Dynatrace Documentation

dt_otel_scope_metrics

Wrap Up#

What You Learned Today#

By completing this lab, you've successfully deployed the OpenTelemetry Collector to collect metrics, enrich metric attributes for better context, and ship those metrics to Dynatrace for analysis.

One Community Contrib Distro OpenTelemetry Collector was deployed as a DaemonSet, behaving as an Agent running on each Node
- The kubeletstats receiver scrapes metrics from the local kubelet on the Node
- The k8sattributes processor enriches the metrics with Kubernetes attributes that may be missing without it
A second Community Contrib Distro OpenTelemetry Collector was deployed as a Deployment, behaving as a Gateway
- The k8s_cluster receiver queries the Kubernetes cluster API to retrieve metrics
- The k8sattributes processor enriches the metrics with Kubernetes attributes that may be missing without it
- The otlp receiver receives signals that are exported from agents, SDKs, and other Collectors
Metrics produced by the OpenTelemetry SDKs and Agents are exported to the otlp receiver
Dynatrace DQL (via Notebooks) allows you to perform powerful queries and analysis of the metric data

Continue#

In the next section, we'll integrate and apply what we have learned in the OpenTelemetry Capstone.

Continue to OpenTelemetry Capstone

OpenTelemetry Metrics#

Prerequisites#

Collector for Node Metrics#

Deploy OpenTelemetry Collector#

Configure Kubernetes RBAC#

kubeletstats receiver#

k8sattributes Processor#

Collector for Cluster Metrics#

Deploy OpenTelemetry Collector#

k8s_cluster receiver#

Export Application Metrics#

otlp receiver#

Wrap Up#

What You Learned Today#

Continue#

`kubeletstats` receiver#

`k8s_cluster` receiver#

`otlp` receiver#