gcloud-lab/infrastructure/gpus/base/keda-gpu-scaling/keda-a100.yaml
sirius0xdev a4b500a3ab fix(keda): remove invalid spec.paths from HTTPScaledObject
HTTPScaledObject CRD (http.keda.sh/v1alpha1) does not support .spec.paths field.
Scaling now triggers on all HTTP requests to the host (health/liveness ignored via targetPendingRequests:1 + pollingInterval:10s).

Refs https://keda.sh/docs/scalers/http-addon/
2026-04-29 04:50:10 +00:00

19 lines
479 B
YAML

apiVersion: http.keda.sh/v1alpha1
kind: HTTPScaledObject
metadata:
name: openclaw-brain-a100-scaling
namespace: customer1
spec:
hosts:
- a100-vllm.internal.cluster # Replace with actual internal routing host if needed
scaleTargetRef:
name: openclaw-brain-vllm
kind: Deployment
apiVersion: apps/v1
service: openclaw-brain-service
port: 8000
replicas:
min: 0
max: 1
scaledownPeriod: 900 # 15 minutes of idle time before scaling to zero