gcloud-lab/infrastructure/gpus/base/keda-gpu-scaling/keda-vllm.yaml

apiVersion: keda.sh/v1alpha1
kind: ScaledObject
metadata:
  name: rtx6000-scaling
  namespace: customer1
spec:
  scaleTargetRef:
    name: rtx6000-brain-vllm
    kind: Deployment
    apiVersion: apps/v1
  hosts:
    - rtx6000-brain-service.customer1.svc.cluster.local
  replicas:
    min: 0
    max: 1
  targetPendingRequests: 1
  scaledownPeriod: 900\n  pollingInterval: 10\n  cooldownPeriod: 30
Update keda-vllm.yaml 2026-04-29 00:05:09 -04:00			`apiVersion: keda.sh/v1alpha1`
			`kind: ScaledObject`
feat: add KEDA scaling for A100 vLLM - Scale to 0 when idle (saves money) - 15 min cooldown before scale-down - Triggers on 3+ pending HTTP requests - Targets openclaw-brain-vllm deployment 2026-04-22 17:07:05 +00:00			`metadata:`
Update keda-vllm.yaml 2026-04-28 23:58:26 -04:00			`name: rtx6000-scaling`
feat: add KEDA scaling for A100 vLLM - Scale to 0 when idle (saves money) - 15 min cooldown before scale-down - Triggers on 3+ pending HTTP requests - Targets openclaw-brain-vllm deployment 2026-04-22 17:07:05 +00:00			`namespace: customer1`
			`spec:`
			`scaleTargetRef:`
Update keda-vllm.yaml 2026-04-28 23:58:26 -04:00			`name: rtx6000-brain-vllm`
fix(keda): correct HTTPScaledObject for A100 vLLM wake-on-request - Use proper scaleTargetRef structure for http.keda.sh/v1alpha1 - Uncomment keda-vllm.yaml in kustomization - This should allow Flux to reconcile the scaler 2026-04-23 02:55:07 +00:00			`kind: Deployment`
			`apiVersion: apps/v1`
fix(keda): use 'hosts' array instead of 'host' to match installed CRD schema Fixes: .spec.host: field not declared in schema error during Flux dry-run. 2026-04-23 04:07:31 +00:00			`hosts:`
Update keda-vllm.yaml 2026-04-28 23:58:26 -04:00			`- rtx6000-brain-service.customer1.svc.cluster.local`
update keda 2026-04-23 01:37:09 +00:00			`replicas:`
			`min: 0`
			`max: 1`
fix keda 2026-04-23 01:42:03 +00:00			`targetPendingRequests: 1`
feat: optimize RTX6000 vLLM KEDA scaling - Rename deployment to rtx6000-brain-vllm to match scaler target ref - Set deployment replicas to 0 (KEDA controlled) - Add pollingInterval: 10s and cooldownPeriod: 30s for faster response This enables scale-to-zero when idle and quick scale-up on first request. See https://github.com/sirius0xdev/gcloud-lab/tree/master/infrastructure%2Fgpus%2Fbase 2026-04-29 04:12:36 +00:00			`scaledownPeriod: 900\n pollingInterval: 10\n cooldownPeriod: 30`