Autoscaling lifecycle

From empty dashboard to burst node provisioned and pod scheduled. Screenshots are captured at each phase during the GCP integration test. To populate or refresh: run ./hack/integration-test-e2e-gcp.sh from cloudburst-autoscaler, then ./scripts/copy-screenshots-to-landing.sh from cloudburst-autoscaler-ui.

Empty

NodePool created, no pending pods, no NodeClaims. Dashboard shows zero burst resources.

Empty overview
Overview — zero burst resources
Empty resource graph
Resource graph — empty

Pending pod

Unschedulable pod created (resource requests exceed existing capacity). Triggers autoscaling.

Overview with pending pod
Overview — pending pod triggers autoscaling
Pods list with Pending pod
Pods — unschedulable pod

NodeClaim pending

Controller creates NodeClaim, CloudBroker recommends instance. NodeClaim in Pending phase until VM is provisioned.

NodeClaims in Pending
NodeClaims — pending
Graph with NodeClaim
Resource graph — NodeClaim created

Node ready

VM bootstrapped, joined via Tailscale, NodeClaim phase Ready. Node appears in cluster.

NodeClaims Ready
NodeClaims — phase Ready
Nodes with burst node
Nodes — burst node joined

Pod scheduled

Pod scheduled onto burst node, Running. Graph shows full NodePool → NodeClaim → Node → Pod chain.

Overview populated
Overview — populated
Pods Running
Pods — running on burst node
Graph populated
Resource graph — NodePool → NodeClaim → Node → Pod

← All dashboard topics

Explore further

Full page tour → All dashboard → Quickstart →