i've a simple kubernetes cluster with a master and 3 minions. In this scenario, if i run a simple pod of a nginx or a mysql it works properly but, if i change type of KIND into yaml file and i try to run a replicated service, pods will start but i can't access to the service.
this is my yaml file for nginx with 3 replicas:
apiVersion: v1
kind: ReplicationController
metadata:
name: nginx
spec:
replicas: 3
selector:
app: nginx
template:
metadata:
name: nginx
labels:
app: nginx
spec:
containers:
- name: nginx
image: nginx
ports:
- containerPort: 80
this is service yaml config file:
apiVersion: v1
kind: Service
metadata:
labels:
name: nginx
name: nginx
spec:
ports:
- port: 80
selector:
name: nginx
i run it with:
# kubectl create -f nginx-rc.yaml
# kubectl create -f nginx-rc-service.yaml
if i run:
# kubectl get pod,svc,rc -o wide
i see:
NAME READY STATUS RESTARTS AGE NODE
nginx-kgq1s 1/1 Running 0 1m node01
nginx-pomx3 1/1 Running 0 1m node02
nginx-xi54i 1/1 Running 0 1m node03
NAME LABELS SELECTOR IP(S) PORT(S)
kubernetes component=apiserver,provider=kubernetes <none> 10.254.0.1 443/TCP
nginx name=nginx name=nginx 10.254.47.150 80/TCP
CONTROLLER CONTAINER(S) IMAGE(S) SELECTOR REPLICAS
nginx nginx nginx app=nginx 3
i can see description for pod:
Name: nginx-kgq1s
Namespace: default
Image(s): nginx
Node: node01/node01
Labels: app=nginx
Status: Running
Reason:
Message:
IP: 172.17.52.3
Replication Controllers: nginx (3/3 replicas created)
Containers:
nginx:
Image: nginx
State: Running
Started: Thu, 11 Feb 2016 16:28:08 +0100
Ready: True
Restart Count: 0
Conditions:
Type Status
Ready True
Events:
FirstSeen LastSeen Count From SubobjectPath Reason Message
Thu, 11 Feb 2016 16:27:47 +0100 Thu, 11 Feb 2016 16:27:47 +0100 1 {scheduler } scheduled Successfully assigned nginx-kgq1s to node01
Thu, 11 Feb 2016 16:27:57 +0100 Thu, 11 Feb 2016 16:27:57 +0100 1 {kubelet node01} implicitly required container POD pulled Pod container image "gcr.io/google_containers/pause:0.8.0" already present on machine
Thu, 11 Feb 2016 16:28:02 +0100 Thu, 11 Feb 2016 16:28:02 +0100 1 {kubelet node01} implicitly required container POD created Created with docker id bed30a90c6eb
Thu, 11 Feb 2016 16:28:02 +0100 Thu, 11 Feb 2016 16:28:02 +0100 1 {kubelet node01} implicitly required container POD started Started with docker id bed30a90c6eb
Thu, 11 Feb 2016 16:28:07 +0100 Thu, 11 Feb 2016 16:28:07 +0100 1 {kubelet node01} spec.containers{nginx} created Created with docker id 0a5c69cd0481
Thu, 11 Feb 2016 16:28:08 +0100 Thu, 11 Feb 2016 16:28:08 +0100 1 {kubelet node01} spec.containers{nginx} started Started with docker id 0a5c69cd0481
this is what i see if i get description for rc:
Name: nginx
Namespace: default
Image(s): nginx
Selector: app=nginx
Labels: app=nginx
Replicas: 3 current / 3 desired
Pods Status: 3 Running / 0 Waiting / 0 Succeeded / 0 Failed
Events:
FirstSeen LastSeen Count From SubobjectPath Reason Message
Thu, 11 Feb 2016 16:27:47 +0100 Thu, 11 Feb 2016 16:27:47 +0100 1 {replication-controller } successfulCreate Created pod: nginx-kgq1s
Thu, 11 Feb 2016 16:27:47 +0100 Thu, 11 Feb 2016 16:27:47 +0100 1 {replication-controller } successfulCreate Created pod: nginx-pomx3
Thu, 11 Feb 2016 16:27:47 +0100 Thu, 11 Feb 2016 16:27:47 +0100 1 {replication-controller } successfulCreate Created pod: nginx-xi54i
and this is what i see if i get description of service:
Name: nginx
Namespace: default
Labels: name=nginx
Selector: name=nginx
Type: ClusterIP
IP: 10.254.47.150
Port: <unnamed> 80/TCP
Endpoints: <none>
Session Affinity: None
No events.
as i can see, the problem may be that i don't have an ENDPOINT but i don't have any idea how i could solve.
It looks to me like the selector for your service is wrong. It's looking for a label of name: nginx
, but your pods actually have app: nginx
.
Try changing your service file to:
apiVersion: v1
kind: Service
metadata:
labels:
name: nginx
name: nginx
spec:
ports:
- port: 80
selector:
app: nginx
... or change your replication controller template to use name: nginx
instead of app: nginx
as the label. Basically, the labels have to match so that the service knows how to present a unified facade over your pods.
See more on this question at Stackoverflow