Jack Francis jackfrancis

Here's what Cluster Autoscaler will do in your scenario, end‑to‑end. The behavior is entirely driven by the fact that the Kubernetes Node objects still exist (kubelet stops, but the API server retains them) and the Azure cloud provider does not surface VM power state to CA.

How CA classifies the 33 stopped nodes

CA buckets every node every loop in clusterstate.go (updateReadinessStats). The buckets are mutually exclusive:

Agent Workflow: Reinforce `azure` support

Auto-generated by providerize on 2026-02-24 22:56 UTC Repository: https://github.com/ray-project/ray

Context

This repository has cloud provider integrations for: aws, gcp, azure.

Current ranking (best → worst):

KubeRay Federation Proposal — Analysis & Comparative Landscape

Proposal Summary

Issue #4561 by @yuchen-ecnu proposes adding federation capability to KubeRay so that a single logical RayCluster can span multiple Kubernetes clusters. The core motivation:

Fragmented GPUs: Organizations procure GPUs across multiple cloud vendors/AZs. Today these are isolated into separate K8s clusters, preventing a unified Ray cluster.
Operational pain: Users must split datasets, deploy multiple small RayClusters, and manually manage them — causing long-tail performance issues and complexity.
Virtual Kubelet limitations: The common workaround (aggregating via Virtual Kubelet) creates control-plane scalability bottlenecks, especially at scale (e.g., 10K→400K+ cores in an hour).

	Here are my two talks:

	The first one will be a joint session w/ GKE (and fellow SIG Autoscaling TL) describing the OSS CA and our plans to make significant changes in 2026.

	The tl;dr is that GOOG will be investing significantly in the OSS CA to correct a decade of forking and blackbox dev.

	This is directly due to investments that we have made in the SIG to modernize the OSS surface area, release process, E2E tests, etc.

	- https://maintainersummiteu2026.sched.com/event/2EWf1/cluster-autoscaler-evolution-kuba-tuznik-google-jack-francis-microsoft

	{
	"kustomize_substitutions": {
	"AZURE_SUBSCRIPTION_ID": "<sub id here>",
	"AZURE_TENANT_ID": "<tenant id here>",
	"AZURE_CLIENT_SECRET": "<sp pw here>",
	"AZURE_CLIENT_ID": "<sp id here>",
	"AZURE_ENVIRONMENT": "AzurePublicCloud",
	"AZURE_SSH_PUBLIC_KEY_B64": "<ssh public key here>"
	},
	"worker-templates": {

	#!/bin/bash

	if [ -z "$RESOURCE_GROUP" ]; then
	echo "must provide a RESOURCE_GROUP env var"
	exit 1;
	fi

	if [ -z "$REGION" ]; then
	echo "must provide a REGION env var"
	exit 1;

	#!/bin/bash

	if [ -z "$RESOURCE_GROUP" ]; then
	echo "must provide a RESOURCE_GROUP env var"
	exit 1;
	fi

	# Continually look for non-Succeeded VMSS instances
	while true; do
	NUM_VMSS=0

	#!/bin/bash

	if [ -z "$GITHUB_USERNAME" ]; then
	echo "Must set the GITHUB_USERNAME variable"
	exit 1
	fi
	if [ -z "$GITHUB_REPOS" ]; then
	# You may set up a bash string array of github repositories for which your GITHUB_USERNAME has a fork, e.g.:
	export GITHUB_REPOS="https://github.com/kubernetes-sigs/cluster-api-provider-azure.git https://github.com/Azure/aks-engine.git \
	https://github.com/weaveworks/kured.git https://github.com/kubernetes-sigs/cluster-api.git https://github.com/kubernetes/perf-tests.git \

	Mar 10 02:12:15 k8s-master-30342830-0 kubelet[16916]: I0310 02:12:15.416617 16916 flags.go:59] FLAG: --add-dir-header="false"
	Mar 10 02:12:15 k8s-master-30342830-0 kubelet[16916]: I0310 02:12:15.417643 16916 flags.go:59] FLAG: --address="0.0.0.0"
	Mar 10 02:12:15 k8s-master-30342830-0 kubelet[16916]: I0310 02:12:15.417660 16916 flags.go:59] FLAG: --allowed-unsafe-sysctls="[]"
	Mar 10 02:12:15 k8s-master-30342830-0 kubelet[16916]: I0310 02:12:15.417676 16916 flags.go:59] FLAG: --alsologtostderr="false"
	Mar 10 02:12:15 k8s-master-30342830-0 kubelet[16916]: I0310 02:12:15.417683 16916 flags.go:59] FLAG: --anonymous-auth="false"
	Mar 10 02:12:15 k8s-master-30342830-0 kubelet[16916]: I0310 02:12:15.417691 16916 flags.go:59] FLAG: --application-metrics-count-limit="100"
	Mar 10 02:12:15 k8s-master-30342830-0 kubelet[16916]: I0310 02:12:15.417698 16916 flags.go:59] FLAG: --authentication-token-webhook="true"
	Mar 10 02:12:15 k8s-master-30342830-0 kubelet[16916]: I0310 02:12:15.417704 16916 flags.go:59] FL