Skip to main content

​
Overview

By adding Kubernetes-specific attributes to telemetry signals, you can gain better visibility into workloads running in the cluster. Helps in correlating traces, logs and metrics with specific Kubernetes resources, such as pods, deployments, nodes and clusters. Adding resource attributes based on custom kubernetes annotations/labels can enrich the data with user defined aggregations and metadata corresponding to the originating pod that produced the telemetry.

​
Configuration Options

The K8sAttributes action is configured using the odigos.io/v1alpha1.Action CRD with the k8sAttributes configuration section.
actionName string : Allows you to attach a meaningful name to the action for convenience.
  • This field is optional
  • Odigos does not use or assume any meaning from this field
notes string : Allows you to attach notes regarding the action for convenience.
  • This field is optional
  • Odigos does not use or assume any meaning from this field
disabled boolean : Allows you to temporarily disable the action, but keep it saved for future use.
  • This field is optional, and defaults to false
signals string[] : An array with the signals that the action will operate on.
  • This field is required
  • Supported values: TRACES, LOGS, METRICS
k8sAttributes object : Configuration for the K8sAttributes action.
  • This field is required for this action type
collectContainerAttributes boolean : Collect the following container related attributes:
  • k8s.container.name
  • container.id
  • container.image.name
  • container.image.tag
This field is optional and defaults to false
collectReplicaSetAttributes boolean : Collect the following replica set related attributes:
  • k8s.replicaset.name if CollectWorkloadUID is set to true, the following attribute will be collected as well:
  • k8s.replicaset.uid
This field is optional and defaults to false
CollectWorkloadUID boolean : Collect the workload UID:
  • k8s.deployment.uid
  • k8s.daemonset.uid
  • k8s.statefulset.uid
Only one of these attributes can appear for a given span/log/metric - based on the workload it originated from. This field is optional and defaults to false
CollectClusterUID boolean : Collect the k8s.cluster.uid attribute, which is set to the uid of the namespace “kube-system”. This allows aggregating observability based on the cluster it originated from.This field is optional and defaults to false
labelsAttributes K8sLabelAttribute[] : Collect resource attributes based on the specified labels.Each label attribute can specify:
  • labelKey: The Kubernetes label key to extract
  • attributeKey: The OpenTelemetry attribute key to create
  • from (deprecated): Single source - “pod” or “namespace”
  • fromSources: Array of sources to extract from, with precedence (pod > namespace > node)
When using fromSources with multiple sources, if the same label exists in multiple sources, the most specific source (e.g., pod) takes precedence over less specific sources (e.g., namespace).This field is optional and can remain empty.
annotationsAttributes K8sAnnotationAttribute[] : Collect resource attributes based on the specified annotations.Each annotation attribute can specify:
  • annotationKey: The Kubernetes annotation key to extract
  • attributeKey: The OpenTelemetry attribute key to create
  • from (deprecated): Single source - “pod” or “namespace”
  • fromSources: Array of sources to extract from, with precedence (pod > namespace > node)
When using fromSources with multiple sources, if the same annotation exists in multiple sources, the most specific source (e.g., pod) takes precedence.This field is optional and can remain empty.

​
Basic Example

The following example demonstrates how to collect Kubernetes attributes, including labels and annotations, and attach them to telemetry signals using the new Action CRD.
1
Create a YAML file with the following content:
k8s-attributes.yaml
apiVersion: odigos.io/v1alpha1
kind: Action
metadata:
  name: collect-k8s-attributes
  namespace: odigos-system
spec:
  actionName: "Collect Kubernetes Attributes"
  signals:
    - TRACES
    - METRICS
    - LOGS
  k8sAttributes:
    collectContainerAttributes: true
    collectWorkloadUID: true
    collectClusterUID: true
    labelsAttributes:
      - labelKey: "app.kubernetes.io/name"
        attributeKey: "app.kubernetes.name"
        from: "pod"
    annotationsAttributes:
      - annotationKey: "kubectl.kubernetes.io/restartedAt"
        attributeKey: "kubectl.kubernetes.restartedAt"
        from: "pod"
2
Apply the action to the cluster:
kubectl apply -f k8s-attributes.yaml

​
Multi-Source Label Extraction Example

The following example demonstrates how to extract the same label from multiple sources with proper precedence. This is useful when you want pod-level labels to override namespace-level labels when both exist.
1
Create a YAML file with the following content:
k8s-attributes-multi-source.yaml
apiVersion: odigos.io/v1alpha1
kind: Action
metadata:
  name: collect-environment-label
  namespace: odigos-system
spec:
  actionName: "Collect Environment Label with Precedence"
  signals:
    - TRACES
    - METRICS
    - LOGS
  k8sAttributes:
    labelsAttributes:
      - labelKey: "environment"
        attributeKey: "k8s.environment"
        fromSources:
          - pod        # Highest precedence - pod labels override namespace labels
          - namespace  # Lower precedence - used as fallback
In this configuration:
  • If a pod has an environment label, that value will be used
  • If the pod doesn’t have the label, but the namespace does, the namespace value will be used
  • This allows namespace-wide defaults while permitting pod-level overrides
2
Apply the action to the cluster:
kubectl apply -f k8s-attributes-multi-source.yaml