kubernetes-vagrant-coreos-cluster
kubernetes-vagrant-coreos-cluster copied to clipboard
no iptables rules being written on host machine
created a test deployment in the kubernetes cluster:
kubectl run hello-world --replicas=2 --labels="run=load-balancer-example" --image=gcr.io/google-samples/node-hello:1.0 --port=8080
I create a service exposing these services:
{
"kind": "Service",
"apiVersion": "v1",
"metadata": {
"name": "example-service",
"namespace": "default",
"selfLink": "/api/v1/namespaces/default/services/example-service",
"uid": "920d2302-c31c-11e7-be6d-0800276974d0",
"resourceVersion": "194855",
"creationTimestamp": "2017-11-06T18:01:53Z",
"labels": {
"pod-template-hash": "2822484588",
"run": "load-balancer-example"
}
},
"spec": {
"ports": [
{
"protocol": "TCP",
"port": 8080,
"targetPort": 8080,
"nodePort": 31561
}
],
"selector": {
"pod-template-hash": "2822484588",
"run": "load-balancer-example"
},
"clusterIP": "10.100.155.72",
"type": "LoadBalancer",
"sessionAffinity": "None",
"externalTrafficPolicy": "Cluster"
},
"status": {
"loadBalancer": {}
}
}
The service starts:
$kubectl describe services example-service
Name: example-service
Namespace: default
Labels: pod-template-hash=2822484588
run=load-balancer-example
Annotations: <none>
Selector: pod-template-hash=2822484588,run=load-balancer-example
Type: LoadBalancer
IP: 10.100.155.72
Port: <unset> 8080/TCP
NodePort: <unset> 31561/TCP
Endpoints: 10.244.48.2:8080,10.244.95.2:8080
Session Affinity: None
Events: <none>
If I ssh inside the vagrant VMs, I can hit the endpoints:
$curl 10.244.95.2:8080
Hello Kubernetes!
On the host, however this give a 'no route to host' error. When I inspect the iptables-save in the host, the reason is clear: no iptables being written on the host
sudo iptables-save | grep example-service
(no results)
I cannot tell if this is a bug or is simply unimplemented?