Kuma — Rate Limit

With this mesh service

$ kubectl get svc -n kuma-demo echo 
NAME   TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)   AGE
echo   ClusterIP   10.96.201.208   <none>        80/TCP    106m

$ kubectl get deployments.apps -n kuma-demo echo 
NAME   READY   UP-TO-DATE   AVAILABLE   AGE
echo   1/1     1            1           112m

And this Rate Limit policy

apiVersion: kuma.io/v1alpha1
kind: RateLimit
mesh: default
metadata:
  name: rl-echo-kuma-demo
spec:
  sources:
    - match:
        kuma.io/service: '*'
  destinations:
    - match:
        kuma.io/service: 'echo_kuma-demo_svc_80'
  conf:
    http:
      requests: 5
      interval: 10s
      onRateLimit:
        status: 429
        headers:
          - key: "do-not-bomb-me"
            value: "true"
            append: true

Sending 2 tps load to echo_kuma-demo_svc_80.mesh.

Result:

$ kubectl exec -it -n kuma-demo2 deployments/nginx -- bash
root@nginx-65df8bc6c6-qrzgk:/# while true; do curl -sS -o /dev/null -v echo_kuma-demo_svc_80.mesh 2>&1|grep -E "^<"|head -1; sleep 0.5; done
< HTTP/1.1 200 OK
< HTTP/1.1 200 OK
< HTTP/1.1 200 OK
< HTTP/1.1 200 OK
< HTTP/1.1 200 OK
< HTTP/1.1 200 OK
< HTTP/1.1 200 OK
< HTTP/1.1 200 OK
< HTTP/1.1 200 OK
< HTTP/1.1 429 Too Many Requests
< HTTP/1.1 429 Too Many Requests
< HTTP/1.1 429 Too Many Requests
< HTTP/1.1 429 Too Many Requests
< HTTP/1.1 429 Too Many Requests
< HTTP/1.1 429 Too Many Requests
< HTTP/1.1 429 Too Many Requests
< HTTP/1.1 429 Too Many Requests
< HTTP/1.1 429 Too Many Requests
< HTTP/1.1 429 Too Many Requests
< HTTP/1.1 429 Too Many Requests
< HTTP/1.1 429 Too Many Requests
< HTTP/1.1 429 Too Many Requests
< HTTP/1.1 429 Too Many Requests
< HTTP/1.1 200 OK
< HTTP/1.1 200 OK

The Rate Limit policy is applied per pod/replica. This means that if a service backend has 3 replicas rate limited to 100 requests per second, the overall service is rate limited to 300 requests per second.

In this example, deploy/echo have only one pod.

refer: https://kuma.io/docs/2.2.x/policies/rate-limit/

That is all.