Kubernetes endpoints throw `ServiceUnavailable`

Question

I have a new Kubernetes cluster on AWS that was built using the kube-up script from v1.1.1. I can successfully access the Elasticsearch/Kibana/KubeUI/Grafana endpoints, but cannot access Heapster/KubeDNS/InfluxDB from my machine, through the API proxy. I have seen some ancillary issues related to this on the K8S project, but no clear identification as to what's going on. From what I can gather, everything is running fine so I'm not sure what is wrong here? I'd really like to use the embedded monitoring of Grafana/Influx/Heapster but the Grafana dashboard is just blank with an series error.

Kubernetes version

$ kubectl version
Client Version: version.Info{Major:"1", Minor:"1", GitVersion:"v1.1.1", GitCommit:"92635e23dfafb2ddc828c8ac6c03c7a7205a84d8", GitTreeState:"clean"}
Server Version: version.Info{Major:"1", Minor:"1", GitVersion:"v1.1.1", GitCommit:"92635e23dfafb2ddc828c8ac6c03c7a7205a84d8", GitTreeState:"clean"}

Cluster-info

$ kubectl cluster-info
Kubernetes master is running at https://MASTER_IP
Elasticsearch is running at https://MASTER_IP/api/v1/proxy/namespaces/kube-system/services/elasticsearch-logging
Heapster is running at https://MASTER_IP/api/v1/proxy/namespaces/kube-system/services/heapster
Kibana is running at https://MASTER_IP/api/v1/proxy/namespaces/kube-system/services/kibana-logging
KubeDNS is running at https://MASTER_IP/api/v1/proxy/namespaces/kube-system/services/kube-dns
KubeUI is running at https://MASTER_IP/api/v1/proxy/namespaces/kube-system/services/kube-ui
Grafana is running at https://MASTER_IP/api/v1/proxy/namespaces/kube-system/services/monitoring-grafana
InfluxDB is running at https://MASTER_IP/api/v1/proxy/namespaces/kube-system/services/monitoring-influxdb

Accessing influxDB from the API proxy URL above

{
  "kind": "Status",
  "apiVersion": "v1",
  "metadata": {},
  "status": "Failure",
  "message": "no endpoints available for service \"monitoring-influxdb\"",
  "reason": "ServiceUnavailable",
  "code": 503
}

Endpoint details from the Host

$ curl http://localhost:8080/api/v1/namespaces/kube-system/endpoints/monitoring-influxdb
{
  "kind": "Endpoints",
  "apiVersion": "v1",
  "metadata": {
    "name": "monitoring-influxdb",
    "namespace": "kube-system",
    "selfLink": "/api/v1/namespaces/kube-system/endpoints/monitoring-influxdb",
    "uid": "2f75b259-8a22-11e5-b248-028ff74b9b1b",
    "resourceVersion": "131",
    "creationTimestamp": "2015-11-13T16:18:33Z",
    "labels": {
      "kubernetes.io/cluster-service": "true",
      "kubernetes.io/name": "InfluxDB"
    }
  },
  "subsets": [
    {
      "addresses": [
        {
          "ip": "10.244.1.4",
          "targetRef": {
            "kind": "Pod",
            "namespace": "kube-system",
            "name": "monitoring-influxdb-grafana-v2-n6jx1",
            "uid": "2f31ed90-8a22-11e5-b248-028ff74b9b1b",
            "resourceVersion": "127"
          }
        }
      ],
      "ports": [
        {
          "name": "http",
          "port": 8083,
          "protocol": "TCP"
        },
        {
          "name": "api",
          "port": 8086,
          "protocol": "TCP"
        }
      ]
    }
  ]
}

Querying the service from the Host

$ curl -IL 10.244.1.4:8083
HTTP/1.1 200 OK
Accept-Ranges: bytes
Content-Length: 13751
Content-Type: text/html; charset=utf-8
Last-Modified: Fri, 14 Nov 2014 21:55:58 GMT
Date: Tue, 17 Nov 2015 21:31:48 GMT

Monitoring-InfluxDB Service

$ curl http://localhost:8080/api/v1/namespaces/kube-system/services/monitoring-influxdb
{
  "kind": "Service",
  "apiVersion": "v1",
  "metadata": {
    "name": "monitoring-influxdb",
    "namespace": "kube-system",
    "selfLink": "/api/v1/namespaces/kube-system/services/monitoring-influxdb",
    "uid": "2f715831-8a22-11e5-b248-028ff74b9b1b",
    "resourceVersion": "60",
    "creationTimestamp": "2015-11-13T16:18:33Z",
    "labels": {
      "kubernetes.io/cluster-service": "true",
      "kubernetes.io/name": "InfluxDB"
    }
  },
  "spec": {
    "ports": [
      {
        "name": "http",
        "protocol": "TCP",
        "port": 8083,
        "targetPort": 8083
      },
      {
        "name": "api",
        "protocol": "TCP",
        "port": 8086,
        "targetPort": 8086
      }
    ],
    "selector": {
      "k8s-app": "influxGrafana"
    },
    "clusterIP": "10.0.35.241",
    "type": "ClusterIP",
    "sessionAffinity": "None"
  },
  "status": {
    "loadBalancer": {}
  }
}

Pod Details

$ kubectl describe pod --namespace=kube-system monitoring-influxdb-grafana-v2-n6jx
Name:               monitoring-influxdb-grafana-v2-n6jx1
Namespace:          kube-system
Image(s):           gcr.io/google_containers/heapster_influxdb:v0.4,beta.gcr.io/google_containers/heapster_grafana:v2.1.1
Node:               ip-172-20-0-44.us-west-2.compute.internal/172.20.0.44
Start Time:         Fri, 13 Nov 2015 08:21:36 -0800
Labels:             k8s-app=influxGrafana,kubernetes.io/cluster-service=true,version=v2
Status:             Running
Reason:
Message:
IP:             10.244.1.4
Replication Controllers:    monitoring-influxdb-grafana-v2 (1/1 replicas created)
Containers:
  influxdb:
    Container ID:   docker://564724318ca81d33d6079978d24f78b3c6ff8eb08a9023c845e250eeb888aafd
    Image:      gcr.io/google_containers/heapster_influxdb:v0.4
    Image ID:       docker://8b8118c488e431cc43e7ff9060968d88402cc6c38a6390c4221352403aa7ac1b
    QoS Tier:
      memory:   Guaranteed
      cpu:  Guaranteed
    Limits:
      memory:   200Mi
      cpu:  100m
    Requests:
      memory:       200Mi
      cpu:      100m
    State:      Running
      Started:      Fri, 13 Nov 2015 08:22:55 -0800
    Ready:      True
    Restart Count:  0
    Environment Variables:
  grafana:
    Container ID:   docker://518dea564a0ee014345e9006da6113fb6584ff1ebc6d0cc9609a608abc995f45
    Image:      beta.gcr.io/google_containers/heapster_grafana:v2.1.1
    Image ID:       docker://200e77ba156a5a86879e49667b97afe84dca42b5bb67ab1e06217e6a19c5a6a6
    QoS Tier:
      cpu:  Guaranteed
      memory:   Guaranteed
    Limits:
      memory:   100Mi
      cpu:  100m
    Requests:
      cpu:      100m
      memory:       100Mi
    State:      Running
      Started:      Fri, 13 Nov 2015 08:22:35 -0800
    Ready:      True
    Restart Count:  0
    Environment Variables:
      INFLUXDB_SERVICE_URL:     http://monitoring-influxdb:8086
      GF_AUTH_BASIC_ENABLED:        false
      GF_AUTH_ANONYMOUS_ENABLED:    true
      GF_AUTH_ANONYMOUS_ORG_ROLE:   Admin
      GF_SERVER_ROOT_URL:       /api/v1/proxy/namespaces/kube-system/services/monitoring-grafana/
Conditions:
  Type      Status
  Ready     True
Volumes:
  influxdb-persistent-storage:
    Type:   EmptyDir (a temporary directory that shares a pod's lifetime)
    Medium:
  grafana-persistent-storage:
    Type:   EmptyDir (a temporary directory that shares a pod's lifetime)
    Medium:
  default-token-bo89c:
    Type:   Secret (a secret that should populate this volume)
    SecretName: default-token-bo89c
No events.

Could you run kubectl decribe pod --namespace=kube-system monitoring-influxdb-grafana-v2-n6jx to see the status of the pod (whether it's ready or not)? — Yu-Ju Hong
You said that "Grafana dashboard is just blank with an series error." What is the error and how it looks like? I think by default, the dashboard shows nothing and you have to manually add graphs yourself. This will change in the next kubernetes release. — Yu-Ju Hong
If I load Grafana, and navigate to the Kubernetes Cluster Dashboard, there are many graphs but they are all empty. They all have errors saying InfluxDB Error: Couldn't find series: <type/metric>. I was also trying to access the InfluxDB UI, and that is was throws the 503 above, and what I'm interested in. — smugcloud
8083 is the port for the admin UI for InfluxDB. 8086 is the API port. Assuming this is InfluxDB 0.9, a better health check for InfluxDB is curl -i http://10.244.1.4:8086/ping, which should return the version number in the header. — beckettsean
@beckettsean If i run your command, I get X-Influxdb-Version: InfluxDB v0.8.9 (git: 664b73e) (leveldb: 1.15). What does this tell us? — smugcloud

Tim Hockin Tim Hockin · Accepted Answer · 2015-11-18T16:37:54

Unfortunately those URLs are incomplete. Influx's ports are named, so you need to say which port you want.

https://MASTER_IP/api/v1/proxy/namespaces/kube-system/services/monitoring-influxdb:http or https://MASTER_IP/api/v1/proxy/namespaces/kube-system/services/monitoring-influxdb:api

There's a bug open to give better errors for this.

Kubernetes endpoints throw `ServiceUnavailable`

1 Answers