Elastic Agent is no work

[xlw712]

version: gocd-22.3.0
Here's how I did it



look agent log is below

yaml config

apiVersion: v1
kind: Namespace
metadata:
  name: gocd
  labels:
    name: gocd
---
apiVersion: v1
kind: Service
metadata:
  name: gocd-server
  namespace: gocd
  annotations:
  labels:
    app: gocd
    version: "gocd-22.3.0"
    component: server
spec:
  type: NodePort
  ports:
    - port: 8153
      targetPort: 8153
      protocol: TCP
      name: http
  selector:
    app: gocd
    component: server
---
kind: PersistentVolume
apiVersion: v1
metadata:
  name: gocd-server-nfs-pv
  namespace: gocd
  labels:
    app: gocd
    version: "gocd-22.3.0"
    component: server
    pv: gocd-server-nfs-pv
spec:
  volumeMode: Filesystem
  capacity:
    storage: 10Gi
  accessModes:
    - ReadWriteOnce
  persistentVolumeReclaimPolicy: Recycle
  nfs:
    server: 11.167.133.162
    path: "/data1/gocd-server"
---
kind: PersistentVolumeClaim
apiVersion: v1
metadata:
  name: gocd-server
  namespace: gocd
  labels:
    app: gocd
    version: "gocd-22.3.0"
    component: server
spec:
  accessModes:
    - "ReadWriteOnce"
  volumeMode: Filesystem
  resources:
    requests:
      storage: "2Gi"
---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: gocd-server
  namespace: gocd
  labels:
    app: gocd
    version: "gocd-22.3.0"
    component: server
spec:
  replicas: 1
  strategy:
    type: Recreate
  selector:
    matchLabels:
      app: gocd
      component: server
  template:
    metadata:
      labels:
        app: gocd
        component: server
    spec:
      securityContext:
        runAsUser: 0
        runAsGroup: 0
        fsGroup: 0
        fsGroupChangePolicy: OnRootMismatch
      serviceAccountName: gocd
      volumes:
        - name: config-vol
          configMap:
            name: gocd
        - name: goserver-vol
          persistentVolumeClaim:
            claimName: gocd-server
      containers:
        - name: gocd-server
          image: "gocd/gocd-server:v22.3.0"
          imagePullPolicy: IfNotPresent
          env:
            - name: GOCD_PLUGIN_INSTALL_kubernetes-elastic-agents
              value: http://192.168.1.1:8888/kubernetes-elastic-agent-3.8.2-356.jar
            - name: GOCD_PLUGIN_INSTALL_docker-registry-artifact-plugin
              value: http://192.168.1.1:8888/docker-registry-artifact-plugin-1.3.1-345.jar
          ports:
            - containerPort: 8153
          livenessProbe:
            httpGet:
              path: /go/api/v1/health
              port: 8153
            initialDelaySeconds: 90
            periodSeconds: 15
            failureThreshold: 10
          readinessProbe:
            httpGet:
              path: /go/api/v1/health
              port: 8153
            initialDelaySeconds: 90
            periodSeconds: 15
            failureThreshold: 10
          volumeMounts:
            - name: config-vol
              mountPath: /preconfigure_server.sh
              subPath: preconfigure_server.sh
            - name: goserver-vol
              mountPath: /godata
              subPath: godata
            - name: goserver-vol
              mountPath: /home/go
              subPath: homego
            - name: goserver-vol
              mountPath: /docker-entrypoint.d
              subPath: scripts
          lifecycle:
            postStart:
              exec:
                command:
                  - /bin/bash
                  - /preconfigure_server.sh
          resources:
            {}
      restartPolicy: Always
---
apiVersion: v1
kind: ServiceAccount
metadata:
  name: gocd
  namespace: gocd
  labels:
    version: "gocd-22.3.0"
    app: "gocd"
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: gocd
  namespace: gocd
  labels:
    version: "gocd-22.3.0"
    app: "gocd"
rules:
- apiGroups: [""]
  resources:
  - pods
  - pods/log
  verbs: ["*"]
- apiGroups: [""]
  resources:
  - nodes
  verbs: ["get", "list"]
- apiGroups: [""]
  resources:
  - events
  verbs: ["list", "watch"]
- apiGroups: [""]
  resources:
  - namespaces
  verbs: ["get"]
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: gocd
  namespace: gocd
  labels:
    version: "gocd-22.3.0"
    app: "gocd"
subjects:
- kind: ServiceAccount
  name: gocd
  namespace: gocd
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: gocd
---
apiVersion: v1
kind: ConfigMap
metadata:
  name: gocd
  namespace: gocd
  labels:
    app: gocd
    version: "gocd-22.3.0"
data:
  preconfigure_server.sh: |-
    #!/bin/bash

    SERVICE_ACCOUNT_PATH=/var/run/secrets/kubernetes.io/serviceaccount
    KUBE_TOKEN=$(<${SERVICE_ACCOUNT_PATH}/token)

    while true
    do
      status_code=$(curl 'http://localhost:8153/go/api/v1/health' -o /dev/null -w "%{http_code}")
      if [ $status_code == 200 ]; then
        break
      fi
      sleep 10
    done

    set -e

    echo "checking if server has already been configured" >> /godata/logs/preconfigure.log

    if [ -f /godata/logs/preconfigure_complete.log ]
    then
      echo "Existing server configuration found in cruise-config.xml. Skipping preconfigure_server scripts." >> /godata/logs/preconfigure.log
      exit 0
    fi

    echo "No configuration found in cruise-config.xml. Using default preconfigure_server scripts to configure server" >> /godata/logs/preconfigure.log

    echo "Trying to configure cluster profile." >> /godata/logs/preconfigure.log

    (curl --fail -i 'http://localhost:8153/go/api/admin/elastic/cluster_profiles' \
      -H'Accept: application/vnd.go.cd+json'  \
      -H 'Content-Type: application/json' \
      -X POST -d '{
        "id": "k8-cluster-profile",
        "plugin_id": "cd.go.contrib.elasticagent.kubernetes",
        "properties": [
              {
                "key": "go_server_url",
                "value": "http://gocd-server:8153/go"
            },
            {
                "key": "kubernetes_cluster_url",
                "value": "https://'$KUBERNETES_SERVICE_HOST':'$KUBERNETES_SERVICE_PORT_HTTPS'"
              },
              {
                "key": "namespace",
                "value": "default"
              },
              {
                "key": "security_token",
                "value": "'$KUBE_TOKEN'"
              }
          ]
      }' >> /godata/logs/preconfigure.log)

    echo "Trying to create an elastic profile now." >> /godata/logs/preconfigure.log

    (curl --fail -i 'http://localhost:8153/go/api/elastic/profiles' \
      -H 'Accept: application/vnd.go.cd+json' \
      -H 'Content-Type: application/json' \
      -X POST -d '{
        "id": "demo-app",
        "cluster_profile_id": "k8-cluster-profile",
        "properties": [
          {
            "key": "Image",
            "value": "gocd/gocd-agent-docker-dind:v22.3.0"
          },
          {
            "key": "PodConfiguration",
            "value": "apiVersion: v1\nkind: Pod\nmetadata:\n  name: gocd-agent-{{ POD_POSTFIX }}\n  labels:\n    app: web\nspec:\n  serviceAccountName: default\n  containers:\n    - name: gocd-agent-container-{{ CONTAINER_POSTFIX }}\n      image: gocd/gocd-agent-docker-dind:v22.3.0\n      securityContext:\n        privileged: true"
          },
          {
            "key": "PodSpecType",
            "value": "yaml"
          },
          {
            "key": "Privileged",
            "value": "true"
          }
        ]
      }' >> /godata/logs/preconfigure.log)

    echo "Trying to creating a hello world pipeline." >> /godata/logs/preconfigure.log

    (curl --fail -i 'http://localhost:8153/go/api/admin/pipelines' \
      -H 'Accept: application/vnd.go.cd+json' \
      -H 'Content-Type: application/json' \
      -X POST -d '{ "group": "sample",
                    "pipeline": {
                      "label_template": "${COUNT}",
                      "name": "getting_started_pipeline",
                      "materials": [
                        {
                          "type": "git",
                          "attributes": {
                            "url": "https://github.com/gocd-contrib/getting-started-repo",
                            "shallow_clone": true
                          }
                        }
                      ],
                      "stages": [
                        {
                          "name": "default_stage",
                          "jobs": [
                            {
                              "name": "default_job",
                              "elastic_profile_id": "demo-app",
                              "tasks": [
                                {
                                  "type": "exec",
                                  "attributes": {
                                    "command": "./build"
                                  }
                                }
                              ],
                              "tabs": [
                                {
                                  "name": "Sample",
                                  "path": "my-artifact.html"
                                }
                              ],
                              "artifacts": [
                                {
                                  "type": "build",
                                  "source": "my-artifact.html"
                                }
                              ]
                            }
                          ]
                        }
                      ]
                    }
                  }' >> /godata/logs/preconfigure.log )

    echo "Done preconfiguring the GoCD server" > /godata/logs/preconfigure_complete.log

What should I do

[chadlwilson]

The error should be pretty clear - UnknownHostException: gocd-server: Name or server not known

The agent cannot find that server at that hostname from wherever it is running. If you relied on the "preconfigure" script as shown above without manually editing the cluster profile, you are probably creating agents in a different namespace to your server: https://github.com/gocd/kubernetes-elastic-agents/blob/master/docs/configure_cluster_profile.md

By default, the Helm chart preconfigure sets gocd to create agents in the same namespace as the server, but in the above you seem to have manually changed the server to be in the gocd namespace, but the agent profile is being preconfigured to default namespace.

So either

• change the Go Server URL in the cluster profile to gocd-server.gocd (namespace suffixed DNS) OR
• change the agents to also be created in the gocd namespace in your cluster profile rather than default namespace

[xlw712]

Ok
I will gocd-server namespace change to default is can work , but change go-agent for web interface to gocd is Error, I don't know why
Here's what I did

[chadlwilson]

Instead of setting the namespace in the pod spec, change it in the Cluster Profile settings. (k8-cluster-profile)

[xlw712]

Ok, I'll try. thanks

[xlw712]

By the way, How to customize go-agent,Do you have any documents? very thanks @chadlwilson

[chadlwilson]

Can you clarify what you mean by "customize go-agent"? Do you mean which tools are pre-installed on the agent? or something else?

[xlw712]

Yes, that's right.
I want to make a gocd-agent based on some special images
for example
install customize jdk
install env
...
What should I do ？

[chadlwilson]

Most people just create their own image and push to their own registry, using one of the GoCD agent images as the base (alpine, debian, centos, ubuntu...).

FROM gocd/gocd-agent-alpine-3.16:v22.3.0
USER root
RUN apk add blah # <-- add your stuff here
USER go

[xlw712]

Ok, Thank you very much ,That was a timely reply