Configure Kubernetes readiness probes (Optional)

This page describes how to use Cloud Service Mesh with Google Cloud health check infrastructure to configure Kubernetes probes and readiness gates.

Prerequisites

As a starting point, this guide assumes that you have already:

• Created a GKE cluster and registered it to a fleet.
• Installed the custom resource definitions.

Transparent Healthchecks with CSM

Cloud Service Mesh Gateway API now offers Transparent health checks, that is, Universal HealthChecks integrated with Kubernetes readiness status data to enable traffic flow to the Pods that are marked ready. This capability is available for both HTTP and gRPC services.

Cloud Service Mesh exposes the status of your Kubernetes Pods directly to the Google health check infrastructure, so the same configuration you use for Pod readinessProbes will be passed transparently to the Google load balancing health check infrastructure and the health check state will be aligned with the readiness probe state. This is done using a component, called Transparent Health Checker, which runs on every node that will serve the Pod's health status.

Cloud Service Mesh configures Health check traffic sent from the health check infrastructure source ranges to be sent to a reserved port (7877). This traffic is redirected to the transparent health check server locally on the node. The server responds with a health status matching the current state of the Kubernetes Pod readiness. The traffic flow to Kubernetes Pods is controlled based on Ready status set on each Pod. You can optionally control the "Ready" status with readiness probe and readiness gates values. The transparent health-check will report the status based on a logical operator AND of all Kubernetes conditions.

You can configure Kubernetes probes, readiness gates and other checks that influence the Kubernetes Pod readiness status.

Configure readiness probes for a HTTP service (Optional)

You can optionally add this section to the Deployment for the whereami service HTTP Service Setup.

...
spec:
      containers:
      - name: whereami
        image: gcr.io/google-samples/whereami:v1.2.20
        ports:
        - containerPort: 8080
        readinessProbe:
          initialDelaySeconds: 1
          periodSeconds: 2
          timeoutSeconds: 1
          successThreshold: 1
          failureThreshold: 1
          httpGet:
            host:
            scheme: HTTP
            path: /
            port: 8080
          initialDelaySeconds: 5
          periodSeconds: 5

Configure readiness probes for a gRPC service (Optional)

You can optionally add this section to the Deployment for psm-grpc-server gRPC service setup.

...
spec:
      containers:
      - name: psm-grpc-server
        image: ${IMAGE_NAME}
        imagePullPolicy: Always
        args:
          ${ARGS}
        ports:
          - containerPort: 50051
        readinessProbe:
          initialDelaySeconds: 1
          periodSeconds: 2
          timeoutSeconds: 1
          successThreshold: 1
          failureThreshold: 1
          grpc:
port: 50051