sirius/gcloud-lab
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Merge pull request #54 from sirius0xdev/fix/keda-httpscaledobject

Browse source 
fix(keda): correct HTTPScaledObject for A100 vLLM wake-on-request
This commit is contained in:
sirius0xdev 2026-04-23 00:03:04 -04:00 • committed by GitHub
commit 65b47f25dd
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
1 changed files with 4 additions and 12 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

16
apps/base/customer1/openclaw/keda-vllm.yaml
View file

@ -4,21 +4,13 @@ metadata:
name: openclaw-brain-a100-scaling name: openclaw-brain-a100-scaling
namespace: customer1 namespace: customer1
spec: spec:
# The host header the interceptor should match
# Target to scale + routing info (old style)
scaleTargetRef: scaleTargetRef:
deployment: openclaw-brain-vllm # your Deployment name name: openclaw-brain-vllm
service: openclaw-brain-service # your Service name kind: Deployment
port: 8000 # ← change if your container port is different (e.g. 8080, 11434 for vLLM, etc.) apiVersion: apps/v1
host: openclaw-brain-service.customer1.svc.cluster.local
# Replica settings
replicas: replicas:
min: 0 min: 0
max: 1 max: 1
host: openclaw-brain-service.customer1.svc.cluster.local
# Scaling metric - best for "wake on first request" use case
targetPendingRequests: 1 targetPendingRequests: 1
# Cooldown before scaling down (15 minutes idle)
scaledownPeriod: 900 scaledownPeriod: 900