Load balance traffic

The ingress controller load balances traffic across pods without further configuration, but you can customize its behavior, such as to set a different load balancing algorithm.

Change the load balancing algorithm for a specific service

By default, the ingress controller uses the round-robin algorithm to distribute requests across a service’s pods. To change the algorithm for a specific service:

• Edit the Ingress resource and set the load-balance annotation to your chosen algorithm. Below, we set the service to use the random algorithm:

   apiVersion: networking.k8s.io/v1
   kind: Ingress
   metadata:
     name: example-ingress
     annotations:
       load-balance: random
   spec:
     ingressClassName: haproxy
     rules:
     - host: "example.com"
       http:
         paths:
         - path: /
           pathType: Prefix
           backend:
             service:
               name: example-service
               port:
                 number: 8080

Enable sticky sessions (session persistence)

In some cases, you may need to route all of a client’s requests to the same backend pod, for example if that pod has stored the client’s server-side session, instead of load balancing the client’s requests across multiple pods. This is called sticky sessions. To enable it for a service:

• Edit the Ingress resource and set the cookie-persistence annotation to a unique name, which the ingress controller uses when naming the HTTP cookie it stores in the client’s browser. The cookie will contain the selected pod’s IP address, port, and a secret key.

  apiVersion: networking.k8s.io/v1
  kind: Ingress
  metadata:
    name: example-ingress
    annotations:
      cookie-persistence: "example-session-persistence-cookie"
  spec:
    ingressClassName: haproxy
    rules:
    - host: "example.com"
      http:
        paths:
        - path: /
          pathType: Prefix
          backend:
            service:
              name: example-service
              port:
                number: 8080

Next up