I have the following docker-compose file that works finely:
version: '3'
services:
myfrontend:
image: myregistry.azurecr.io/im1:latest
container_name: myfrontend
ports:
- 80:80
- 443:443
mybackend:
image: myregistry.azurecr.io/im2:latest
container_name: mybackend
expose:
- 8080
The backend only exposes 8080 to the internal network, the frontend has a modded nginx image with the following configuration (and it works as docker resolves the ip with the container name)
server {
listen 80 default_server;
location / {
auth_basic "Restricted";
auth_basic_user_file /etc/nginx/.htpasswd;
resolver 127.0.0.11 ipv6=off;
set $springboot "http://mybackend:8080";
proxy_pass $springboot;
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header Host $host;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
}
I migrated the above configuration into kubernates and I get a 502 bad gateway error from nginx, I think because it cannot solve the backend address.
Here's the kubernates conf, can you give it a look and tell me what am I doing wrong? :(
apiVersion: apps/v1beta1
kind: Deployment
metadata:
name: mybackend
spec:
replicas: 1
strategy:
rollingUpdate:
maxSurge: 1
maxUnavailable: 1
minReadySeconds: 5
template:
metadata:
labels:
app: mybackend
spec:
nodeSelector:
"beta.kubernetes.io/os": linux
containers:
- name: mybackend
image: myregistry.azurecr.io/sgr-mybackend:latest
ports:
- containerPort: 8080
name: mybackend
resources:
requests:
cpu: 250m
limits:
cpu: 500m
---
apiVersion: v1
kind: Service
metadata:
name: mybackend
spec:
ports:
- port: 8080
selector:
app: mybackend
---
apiVersion: apps/v1beta1
kind: Deployment
metadata:
name: myfrontend
spec:
replicas: 1
template:
metadata:
labels:
app: myfrontend
spec:
nodeSelector:
"beta.kubernetes.io/os": linux
containers:
- name: myfrontend
image: myregistry.azurecr.io/myfrontend:latest
ports:
- containerPort: 80
---
apiVersion: v1
kind: Service
metadata:
name: myfrontend
spec:
type: LoadBalancer
ports:
- port: 80
selector:
app: myfrontend
you need to set your resolver to this:
kube-dns.kube-system.svc.cluster.local
so the kube-dns name\address in your cluster, because nothing on localhost would resolve mybackend to its ip address. I'm not sure you need this at all, because container would know backend address from kubernetes anyway. I'd probably drop that setting