L4 Load Balancing

echoserver is a simple service that provide standard internet services like time, daytime, chargen, and standard echo service. echoserver are deployed in k8a as container with multiple replicas, now need a load balancer and ingress controller to let the outside request can goes into echoserver.

Deploy Echoserver

git clone https://github.com/cloudadc/container-ingress.git
cd container-ingress/nginx-plus-ingress/l4-lb

kubectl apply -f https://www.xnip.cn/doc/AxDdDcc8Tfdeploy.yaml

• https://www.xnip.cn/doc/AxDdDcc8Tfdeploy.yaml

this will create a namespace named echoserver, and the echoserver service under it.

$ telnet 192.168.5.30 30165
Trying 192.168.5.30...
Connected to 192.168.5.30.
Escape character is '^]'.
hello
kylin

Deploy Ingress Controller

Either refer to Installing, or Deployment Architectures to choose one for dedetailed instructions to deploy Ingress Controller.

eg: deploy a cluster-wide nginx plus ingress controller:

// 1. deploy cluster wide nginx ingress controller
kubectl apply -f https://www.xnip.cn/doc/AxDdDcc8Tfnginx-plus-ingress.yaml

// 2. deploy bigip controller
kubectl apply -f https://www.xnip.cn/doc/AxDdDcc8Tfnginx-plus-ingress-svc.yaml
kubectl apply -f https://www.xnip.cn/doc/AxDdDcc8Tfnginx-plus-ingress-configmap.yaml

• https://www.xnip.cn/doc/AxDdDcc8Tfnginx-plus-ingress.yaml

• https://www.xnip.cn/doc/AxDdDcc8Tfnginx-plus-ingress-svc.yaml

• https://www.xnip.cn/doc/AxDdDcc8Tfnginx-plus-ingress-configmap.yaml

Check the dashboard via http://192.168.5.40:8080/dashboard.html

Configure Load Balancing

We use stream-snippets ConfigMap key to configure TCP load balancing for our echoserver pods.

kubectl apply -f https://www.xnip.cn/doc/AxDdDcc8Tfnginx-plus-config.yaml

https://www.xnip.cn/doc/AxDdDcc8Tfnginx-plus-config.yaml

      resolver kube-dns.kube-system.svc.cluster.local valid=5s;

      upstream echoserver-tcp {
          zone echoserver-tcp 64k;
          server echoserver-svc.echoserver.svc.cluster.local service=_echoserver._tcp resolve;
      }

      server {
          listen 8877;
          proxy_pass echoserver-tcp;
          status_zone echoserver-tcp;
      }

kubectl exec -it nginx-ingress-7fbffb7bc6-hrq7q -n nginx-ingress -- bash

$ cat /etc/nginx/nginx.conf
...
    upstream echoserver-tcp {
        zone echoserver-tcp 64k;
        server echoserver-svc.echoserver.svc.cluster.local service=_echoserver._tcp resolve;
    }

    server {
        listen 8877;
        proxy_pass echoserver-tcp;
        status_zone echoserver-tcp;
    }

4. Check the Nginx Plus Ingress Controller Dashboard

Note that now has one TCP/UDP zone and one TCP/UDP upstream.

5. Check the Ingress path

From the F5 LTM, there is public IP which used as unified external enterence as below figure, it’s listen on 192.168.5.40:8877

Note that the Virtual Server IP reference to 2 pool member, that because we deploy nginx plus ingress has 2 replicas.

From Nginx Plus TCP/UDP Upstreams, there only one records, the echoserver service ip is used,

Demonstrations

Either use telnet or echoclient:

$ telnet 192.168.5.40 8877
Trying 192.168.5.40...
Connected to 192.168.5.40.
Escape character is '^]'.
hello
hello
kylin
kylin

$ echoclient 192.168.5.40
daytime
Sat Aug 15 20:41:51 2020
time
1597524114
hello
hello

check the statistics from F5

Note that the 2 connection all are established via 1st ingress controller.