chore(perf): pin kube-prometheus-stack to specific chart version and disable migration job

Updated application-kube-prometheus-stack.yaml to pin the targetRevision to kube-prometheus-stack-83.7.0 for stability. Disabled the migration job in litellm-values.yaml to streamline deployment, allowing the proxy image to handle DB setup on startup.

deploy/performance/argocd/application-kube-prometheus-stack.yaml

@@ -1,6 +1,7 @@
 # Argo CD Application: kube-prometheus-stack → namespace monitoring.
 #
 # Chart: https://github.com/prometheus-community/helm-charts/tree/main/charts/kube-prometheus-stack
+# Pinned to upstream chart release tag (not floating `main`). Bump `targetRevision` when upgrading.
 #
 # Apply:
 #   kubectl apply -n argocd -f deploy/performance/argocd/application-kube-prometheus-stack.yaml
@@ -14,7 +15,7 @@ spec:
   project: default
   source:
     repoURL: https://github.com/prometheus-community/helm-charts.git
-    targetRevision: main
+    targetRevision: kube-prometheus-stack-83.7.0
     path: charts/kube-prometheus-stack
     helm:
       releaseName: kube-prometheus-stack

deploy/performance/helm/litellm-values.yaml

@@ -91,10 +91,14 @@ redis:
     persistence:
       storageClass: gp2
 
+# Skip separate Batch Job + Argo PreSync: proxy image runs DB setup on startup when
+# DISABLE_SCHEMA_UPDATE is not set (see litellm/proxy/proxy_cli.py). Use the Job again if you
+# prefer migrations before any proxy pod starts, or if you run many replicas and want a single migration step.
 migrationJob:
+  enabled: false
   hooks:
     argocd:
-      enabled: true
+      enabled: false
 
 proxy_config:
   model_list: