I have on-premise k8s cluster with 3 masters and 4 workers. Metallb is installed and DHCP is provided for him from the same network as on-premise cluster IPs.
I tried to start Kubernetes Dashboard app in the cluster and everything is working fine except LoadBalancer type of Service of Kubernetes Dashboard:
apiVersion: v1
metadata:
annotations:
metallb.universe.tf/address-pool: single-ip
labels:
k8s-app: kubernetes-dashboard
name: kubernetes-dashboard
namespace: kubernetes-dashboard
spec:
type: LoadBalancer
ports:
- port: 443
targetPort: 8443
selector:
k8s-app: kubernetes-dashboard
externalIPs:
- 192.168.1.129
root@master1:/# kubectl get services -n kubernetes-dashboard
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
dashboard-metrics-scraper ClusterIP 10.249.40.230 <none> 8000/TCP 6h8m
kubernetes-dashboard LoadBalancer 10.249.3.22 192.168.1.129 443:32593/TCP 6h8m
After applying all configs dashboard is starting to work at https://192.168.1.129 from cluster and from my computer correctly.
But after some time stopped to work from my computer but continue to work from cluster. From my computer I'm getting "Connection refused error".
From my computer and from cluster 192.168.1.129 IP is pinging normally without any problems.
externalTrafficPolicy is Cluster by default. Changing it to Local didn't help.
What's happening? How to resolve issue when service working and then stopping to work and then again starting to work? :)