add wellness probes to vllm and activate keda

apps/base/customer1/hermes-agent/deployment.yaml

@@ -50,8 +50,9 @@ spec:
           - name: HERMES_MODEL_PROVIDER
             value: xai
           
-          - name: hermes-model
+          - name: HERMES_MODEL 
             value: grok-4.1-fast
+          
           - name: OPENAI_API_KEY
             value: "dummy"                                 # vLLM ignores this
 

apps/base/customer1/openclaw/a100-vllm.yaml

@@ -39,14 +39,17 @@ spec:
             - --tokenizer-mode=hf 
             - --gpu-memory-utilization=0.95
             - --kv-cache-dtype=fp8
-            - --max-model-len=65536
+            - --max-model-len=131072 
             - --enable-auto-tool-choice 
+            - --enable-chunked-prefill
+            - --max-num-batched-tokens=8192
+            - --max-num-seqs=4
             - --trust-remote-code
             - --dtype=auto 
             - --enable-prefix-caching
             - --tool-call-parser=qwen3_xml
             - --reasoning-parser=qwen3
-            
+            - --limit-mm-per-prompt='{"image":1, "video":0 }'
             - --disable-custom-all-reduce
           ports:
             - containerPort: 8000
@@ -59,6 +62,38 @@ spec:
               nvidia.com/gpu: 1
               memory: "80Gi"
               cpu: "8"
+          
+          startupProbe:
+            httpGet:
+              path: /health
+              port: 8000
+            initialDelaySeconds: 30
+            periodSeconds: 10
+            timeoutSeconds: 10
+            failureThreshold: 60
+            successThreshold: 1 
+          
+          livenessProbe:
+            httpGet:
+              path: /health
+              port: 8000 
+            initialDelaySeconds: 180
+            periodSeconds: 15
+            failureThreshold: 5
+            successThreshold: 1 
+            timeoutSeconds: 10 
+            
+          
+          readinessProbe:
+            httpGet:
+              path: /v1/models 
+              port: 8000
+            initialDelaySeconds: 120
+            periodSeconds: 10 
+            timeoutSeconds: 15 
+            failureThreshold: 8
+            successThreshold: 1 
+          
           volumeMounts:
             - name: model-cache
               mountPath: /root/.cache/huggingface

apps/base/customer1/openclaw/kustomization.yaml

@@ -2,12 +2,12 @@ apiVersion: kustomize.config.k8s.io/v1beta1
 kind: Kustomization
 resources:
  # - vllm-l4.yaml
- # - keda-a100.yaml
+  - keda-a100.yaml
   - pvc.yaml
   - new-configmap.yaml
   - deployment.yaml
   - service.yaml
   - openclaw-secrets.yaml
   - pg-cluster-openclaw.yaml
- # - a100-vllm.yaml 
+  - a100-vllm.yaml 
   - xai-apikey.yaml