Cette page n'est pas encore disponible en français, sa traduction est en cours.
Si vous avez des questions ou des retours sur notre projet de traduction actuel, n'hésitez pas à nous contacter.

This document contains troubleshooting information for the following components:

Datadog Cluster Agent

To execute the troubleshooting commands for the Cluster Agent, you first need to be inside the Cluster Agent or the node-based Agent pod. For this, use:

kubectl exec -it <DATADOG_CLUSTER_AGENT_POD_NAME> bash

To see what cluster level metadata is served by the Datadog Cluster Agent, run:

agent metamap

You should see the following result:

root@datadog-cluster-agent-8568545574-x9tc9:/# agent metamap

Metadata Mapper

Node detected: gke-test-default-pool-068cb9c0-sf1w

  - Namespace: kube-system
      - Pod: kube-dns-788979dc8f-hzbj5
        Services: [kube-dns]
      - Pod: kube-state-metrics-5587867c9f-xllnm
        Services: [kube-state-metrics]
      - Pod: kubernetes-dashboard-598d75cb96-5khmj
        Services: [kubernetes-dashboard]

Node detected: gke-test-default-pool-068cb9c0-wntj

  - Namespace: default
      - Pod: datadog-cluster-agent-8568545574-x9tc9
        Services: [datadog-custom-metrics-server dca]

  - Namespace: kube-system
      - Pod: heapster-v1.5.2-6d59ff54cf-g7q4h
        Services: [heapster]
      - Pod: kube-dns-788979dc8f-q9qkt
        Services: [kube-dns]
      - Pod: l7-default-backend-5d5b9874d5-b2lts
        Services: [default-http-backend]
      - Pod: metrics-server-v0.2.1-7486f5bd67-v827f
        Services: [metrics-server]

To verify that the Datadog Cluster Agent is being queried, look for:

root@datadog-cluster-agent-8568545574-x9tc9:/# tail -f /var/log/datadog/cluster-agent.log
2018-06-11 09:37:20 UTC | DEBUG | (metadata.go:40 in GetPodMetadataNames) | CacheKey: agent/KubernetesMetadataMapping/ip-192-168-226-77.ec2.internal, with 1 services
2018-06-11 09:37:20 UTC | DEBUG | (metadata.go:40 in GetPodMetadataNames) | CacheKey: agent/KubernetesMetadataMapping/ip-192-168-226-77.ec2.internal, with 1 services

If you are not collecting events properly, ensure that DD_LEADER_ELECTION and DD_COLLECT_KUBERNETES_EVENTS are set to true, as well as the proper verbs listed in the RBAC (notably, watch events).

If you have enabled those, check the leader election status and the kube_apiserver check with the following command:

agent status

This should produce the following result:

root@datadog-cluster-agent-8568545574-x9tc9:/# agent status
  Leader Election
    Leader Election Status:  Running
    Leader Name is: datadog-cluster-agent-8568545574-x9tc9
    Last Acquisition of the lease: Mon, 11 Jun 2018 06:38:53 UTC
    Renewed leadership: Mon, 11 Jun 2018 09:41:34 UTC
    Number of leader transitions: 2 transitions
  Running Checks
      Total Runs: 736
      Metrics: 0, Total Metrics: 0
      Events: 0, Total Events: 100
      Service Checks: 3, Total Service Checks: 2193

Node Agent

You can check the status of the Datadog Cluster Agent by running the Agent status command: agent status

If the Datadog Cluster Agent is enabled and correctly configured, you should see:

 Datadog Cluster Agent
   - Datadog Cluster Agent endpoint detected: https://XXX.XXX.XXX.XXX:5005
   Successfully Connected to the Datadog Cluster Agent.
   - Running: {Major:1 Minor:0 Pre:xxx Meta:xxx Commit:xxxxx}

Make sure the Cluster Agent service was created before the Agents’ Pods, so that the DNS is available in the environment variables:

root@datadog-agent-9d5bl:/# env | grep DATADOG_CLUSTER_AGENT | sort

root@datadog-agent-9d5bl:/# echo ${DD_CLUSTER_AGENT_AUTH_TOKEN}

Further Reading