Skip to main content

Exposing socket io and webservice on a service in EKS, ALB

Lets say one of your microservices has both socketio and regular http service endpoints too, so how do you expose both services on EKS using AWS ALB?

Can be tricky at first because you need to make the service available via your general api domain by registering the backend in your ingress.

alb.yaml
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
name: My Ingress
annotations:
kubernetes.io/ingress.class: alb
spec:
rules:
- http:
paths:
- path: /v1/driver/
pathType: Prefix
backend:
service:
name: driver
port:
number: 80

At the same time you want to you want the expose the service itself on the other hand via LoadBalancer instead of using NodePort, likewise you want to make sure your healthcheck is in place to avoid spammy logs from aws target.

At first i thought it wont be possible to have both ingress and service annotation in a single deployment file, but it turns out to be true, jaw breaking to see it work πŸ˜‚.

Here is what your final serive yaml file will look like.

alb.yaml
apiVersion: v1
kind: Service
metadata:
annotations:
service.beta.kubernetes.io/aws-load-balancer-healthcheck-protocol: HTTP
service.beta.kubernetes.io/aws-load-balancer-healthcheck-port: 'traffic-port'
service.beta.kubernetes.io/aws-load-balancer-healthcheck-path: /v1/drivers/alive
service.beta.kubernetes.io/aws-load-balancer-healthcheck-success-codes: "200"
service.beta.kubernetes.io/aws-load-balancer-healthcheck-timeout: '5'
service.beta.kubernetes.io/aws-load-balancer-healthcheck-healthy-threshold: '2'
service.beta.kubernetes.io/aws-load-balancer-healthcheck-unhealthy-threshold: '2'
alb.ingress.kubernetes.io/healthcheck-protocol: HTTP
alb.ingress.kubernetes.io/healthcheck-port: traffic-port
alb.ingress.kubernetes.io/healthcheck-path: /v1/drivers/alive
alb.ingress.kubernetes.io/healthcheck-timeout-seconds: '5'
alb.ingress.kubernetes.io/success-codes: '200'
alb.ingress.kubernetes.io/healthy-threshold-count: '2'
alb.ingress.kubernetes.io/unhealthy-threshold-count: '2'
name: drivers
spec:
selector:
app: drivers
ports:
- protocol: TCP
port: 80
type: LoadBalancer

You will endup have two target group for a single service, one for the socketio exposed via LoadBalancer and one for the Ingress Backend.

its not like you cant still access your HTTP endpoints from the loadbalancer too, but since you are dealing with a microservice here, thats the best way i could think off.

OFcourse if you are using a monolithic service, you are good to go with only service.beta.kubernetes.io annotation.

Wondering why i didnt use the TCP health check also? that thing just get your logs spammed with unrelated logs.

Chill

Till next time ✌️


Comments