ApisixClusterConfig

ApisixClusterConfig is a CRD resource which used to describe an APISIX cluster, currently it's not a required resource but its existence can enrich an APISIX cluster, for instance, enabling cluster-wide monitoring, rate limiting and so on.

monitoring features like collecting Prometheus metrics and skywalking spans

Monitoring

By default, monitoring are not enabled for the APISIX cluster, this is not favorable if you'd like to learn the real running status of your cluster. In such a case, you could create a ApisixClusterConfig to enable these features explicitly.

apiVersion: apisix.apache.org/v2
kind: ApisixClusterConfig
metadata:
  name: default
spec:
  monitoring:
    prometheus:
      enable: true
    skywalking:
      enable: true
      sampleRatio: 0.5

The above example enables both the Prometheus and Skywalking for the APISIX cluster which name is "default". Please see Prometheus in APISIX and Skywalking in APISIX for the details.

Admin Config

The default APISIX cluster is configured through command line options like --default-apisix-cluster-xxx. They are constant in apisix-ingress-controller's lifecycle, you have to change the definition of Deployment or Pod template. Now with the help of ApisixClusterConfig, you can change some administrative fields on it.

apiVersion: apisix.apache.org/v2
kind: ApisixClusterConfig
metadata:
  name: default
spec:
  admin:
    baseURL: http://apisix-gw.default.svc.cluster.local:9180/apisix/admin
    adminKey: "123456"

The above ApisixClusterConfig sets the base url and admin key for the APISIX cluster "default". Once this resource is processed, resources like Route, Upstream and others will be pushed to the new address with the new admin key (for authentication).

Multiple Clusters Management

ApisixClusterConfig is also designed for supporting multiple clusters management, but currently this function IS NOT ENABLED YET. Only the ApisixClusterConfig with the same named configured in --default-apisix-cluster-name option will be handled by apisix-ingress-controller, other instances will be neglected.

The current delete event for ApisixClusterConfig doesn't mean the apisix-ingress-controller will lose the view of the corresponding APISIX cluster but resetting all the features on it, so the running of APISIX cluster is not influenced by this event.