Provide a quick view that shows the health of a load balancer in a GKE app

[BSick7]

Recently, we attached a load balancer to a GKE app.

There are several ways that it could fail and it was difficult to track down:

• Is the gateway launched/healthy?
• Is the SSL provisioned properly?
• Is the health check correct?
• Is the port correct?

There is information that sprawls several gcp and k8s to diagnose. Here is a checklist:

1. kubectl describe gateway + kubectl describe httproute (Accepted/Programmed/ResolvedRefs)
2. kubectl get endpointslice (does the Service actually have endpoints?)
3. gcloud compute backend-services get-health … (why unhealthy)
4. gcloud compute health-checks describe … (what it’s probing)
5. gcloud compute network-endpoint-groups list-network-endpoints … (are endpoints registered?)

Ironically, I only found the solution by navigating to the pod in GCP console.

Some interesting notes:

• If I click on the replica set, there is no useful information
• If I click on the pod revision, there is no useful information -- it navigates to the replica set (wtf?)
• I had to click on the new pod name to find the information under Events tab. (Revealed a readiness probe issue)