use prometheus with external ip address - kubernetes

Question

we have k8s cluster and I’ve application which is running there.
Now I try to add https://prometheus.io/
and I use the command
helm install stable/prometheus --version 6.7.4 --name my-prometheus
this command works and I got this
NAME: my-prometheus
LAST DEPLOYED: Tue Feb 5 15:21:46 2019
NAMESPACE: default
STATUS: DEPLOYED
...
when I run command
kubectl get services
I got this
kubernetes ClusterIP 100.64.0.1 <none> 443/TCP 2d4h
my-prometheus-alertmanager ClusterIP 100.75.244.55 <none> 80/TCP 8m44s
my-prometheus-kube-state-metrics ClusterIP None <none> 80/TCP 8m43s
my-prometheus-node-exporter ClusterIP None <none> 9100/TCP 8m43s
my-prometheus-pushgateway ClusterIP 100.75.24.67 <none> 9091/TCP 8m43s
my-prometheus-server ClusterIP 100.33.26.206 <none> 80/TCP 8m43s
I didnt get any externalIP
Does someone knows how to add it ? via service? any example for this
update
i’ve added the following yml
apiVersion: v1
kind: Service
metadata:
name: prometheus-service
spec:
selector:
app: prometheus-server
type: LoadBalancer
ports:
- port: 8080
targetPort: 9090
nodePort: 30001
which created successfully
now I see the external ip like when running kubectl get services
my-prometheus-server LoadBalancer 100.33.26.206 8080:30001/TCP 80/TCP 8m43s
And I use in the browser 100.33.26.206:30001 and nothing happen, any idea?

Answer 1

I think what you are trying to do is to create a service with a type LoadBalancer, those have an internal and external IP.
You can create one like any other service but you should precise those two fields:
externalTrafficPolicy: Local
type: LoadBalancer
Updated:
There seems to be some confusion, you don't need an external ip to monitor your apps, it will only be used to access prometheus UI.
The UI is accessible on port 9090 but prometheus is never accessed by the exporter as it will be prometheus wich will be scraping the exporters.
Now to access a service from the internet you should have a google ip, but it seems that what you have is still an internal IP, it's in the same subnet as the other clusterIP, and it should not. For now in place of an external ip it's showing a port redirect wich is also wrong as the prometheus UI is on port 9090 (if you didn't modify your configuration it should still be). You should try to remove the "nodePort" and leave the port redirect to kubernetes.

Answer 2

The Prometheus helm chart does support configuration for service, see the documentation
To configure Prometheus server on a local cluster, follow the steps:
Create values.yaml:
server:
service:
servicePort: 31000
type: LoadBalancer
loadBalancerIP: localhost
or
server:
service:
nodePort: 31000
type: NodePort
Add stable repo to helm (if missing):
helm repo add stable "https://kubernetes-charts.storage.googleapis.com/"
Install Prometheus:
helm install prometheus-demo stable/prometheus --values .\values.yaml
Wait for 1-2mins. Prometheus should be available: http://localhost:31000/