right-size 14 services and scale down GPU-heavy workloads [ci skip]

Memory right-sizing based on VPA upperBound analysis:
- Increases: stirling-pdf 1200→1536Mi, claude-memory 64→128Mi,
  dawarich 512→768Mi, kyverno-cleanup 128→192Mi, linkwarden 768→1Gi,
  navidrome 64→128Mi, listenarr 768→896Mi, privatebin 64→128Mi,
  ntfy 64→128Mi, health 128→256Mi, dbaas quota 16→20Gi,
  mysql-operator 384→512Mi
- Decreases: rybbit 768→384Mi, nvidia-exporter added explicit 192Mi,
  dcgm-exporter 2560→1536Mi
- Scale to 0: ebook2audiobook/audiblez-web, whisper (GPU node pressure)

Net effect: -496Mi cluster-wide, 13 ContainerNearOOM alerts resolved,
all ResourceQuota pressures cleared, GPU health green.

stacks/claude-memory/main.tf

@@ -224,11 +224,11 @@ resource "kubernetes_deployment" "claude-memory" {
 
           resources {
             requests = {
-              memory = "64Mi"
+              memory = "128Mi"
               cpu    = "10m"
             }
             limits = {
-              memory = "64Mi"
+              memory = "128Mi"
             }
           }
         }

stacks/dawarich/main.tf

@@ -176,10 +176,10 @@ resource "kubernetes_deployment" "dawarich" {
           resources {
             requests = {
               cpu    = "15m"
-              memory = "512Mi"
+              memory = "768Mi"
             }
             limits = {
-              memory = "512Mi"
+              memory = "768Mi"
             }
           }
         }

stacks/ebook2audiobook/main.tf

@@ -326,7 +326,7 @@ resource "kubernetes_deployment" "audiblez-web" {
     }
   }
   spec {
-    replicas = 1
+    replicas = 0 # Scaled down - GPU node memory pressure
     strategy {
       type = "Recreate"
     }

stacks/health/main.tf

@@ -107,11 +107,11 @@ resource "kubernetes_deployment" "health" {
 
           resources {
             requests = {
-              memory = "128Mi"
+              memory = "256Mi"
               cpu    = "15m"
             }
             limits = {
-              memory = "128Mi"
+              memory = "256Mi"
             }
           }
         }

stacks/linkwarden/main.tf

@@ -182,10 +182,10 @@ resource "kubernetes_deployment" "linkwarden" {
           resources {
             requests = {
               cpu    = "50m"
-              memory = "768Mi"
+              memory = "1Gi"
             }
             limits = {
-              memory = "768Mi"
+              memory = "1Gi"
             }
           }
         }

stacks/navidrome/main.tf

@@ -133,10 +133,10 @@ resource "kubernetes_deployment" "navidrome" {
           resources {
             requests = {
               cpu    = "15m"
-              memory = "64Mi"
+              memory = "128Mi"
             }
             limits = {
-              memory = "64Mi"
+              memory = "128Mi"
             }
           }
         }

stacks/ntfy/main.tf

@@ -121,10 +121,10 @@ resource "kubernetes_deployment" "ntfy" {
           resources {
             requests = {
               cpu    = "10m"
-              memory = "64Mi"
+              memory = "128Mi"
             }
             limits = {
-              memory = "64Mi"
+              memory = "128Mi"
             }
           }
         }

stacks/platform/modules/dbaas/main.tf

@@ -35,8 +35,8 @@ resource "kubernetes_resource_quota" "dbaas" {
   spec {
     hard = {
       "requests.cpu"    = "8"
-      "requests.memory" = "16Gi"
+      "requests.memory" = "20Gi"
-      "limits.memory"   = "16Gi"
+      "limits.memory"   = "20Gi"
       pods              = "30"
     }
   }
@@ -78,10 +78,10 @@ resource "helm_release" "mysql_operator" {
     resources = {
       requests = {
         cpu    = "100m"
-        memory = "384Mi"
+        memory = "512Mi"
       }
       limits = {
-        memory = "384Mi"
+        memory = "512Mi"
       }
     }
   })]

stacks/platform/modules/kyverno/main.tf

@@ -57,11 +57,11 @@ resource "helm_release" "kyverno" {
     cleanupController = {
       resources = {
         limits = {
-          memory = "128Mi"
+          memory = "192Mi"
         }
         requests = {
           cpu    = "100m"
-          memory = "128Mi"
+          memory = "192Mi"
         }
       }
     }

stacks/platform/modules/nvidia/main.tf

@@ -132,7 +132,11 @@ resource "kubernetes_deployment" "nvidia-exporter" {
             }
           }
           resources {
+            requests = {
+              memory = "192Mi"
+            }
             limits = {
+              memory           = "192Mi"
               "nvidia.com/gpu" = "1"
             }
           }

stacks/platform/modules/nvidia/values.yaml

@@ -18,13 +18,13 @@ driver:
     config:
       name: time-slicing-config
 
-# DCGM Exporter - increased from 768Mi to 2560Mi based on VPA upper bound of 2091Mi
+# DCGM Exporter - reduced from 2560Mi to 1536Mi based on VPA upper bound of 1459Mi (1.05x margin)
 dcgmExporter:
   resources:
     requests:
-      memory: "2560Mi"
+      memory: "1536Mi"
     limits:
-      memory: "2560Mi"
+      memory: "1536Mi"
 
 # CUDA Validator - reduced from 1024Mi to 256Mi (one-shot job)
 validator:

stacks/privatebin/main.tf

@@ -70,10 +70,10 @@ resource "kubernetes_deployment" "privatebin" {
           resources {
             requests = {
               cpu    = "10m"
-              memory = "64Mi"
+              memory = "128Mi"
             }
             limits = {
-              memory = "64Mi"
+              memory = "128Mi"
             }
           }
         }

stacks/rybbit/main.tf

@@ -353,10 +353,10 @@ resource "kubernetes_deployment" "rybbit" {
           resources {
             requests = {
               cpu    = "25m"
-              memory = "768Mi"
+              memory = "384Mi"
             }
             limits = {
-              memory = "768Mi"
+              memory = "384Mi"
             }
           }
         }

stacks/servarr/listenarr/main.tf

@@ -59,10 +59,10 @@ resource "kubernetes_deployment" "listenarr" {
           resources {
             requests = {
               cpu    = "25m"
-              memory = "768Mi"
+              memory = "896Mi"
             }
             limits = {
-              memory = "768Mi"
+              memory = "896Mi"
             }
           }
         }

stacks/stirling-pdf/main.tf

@@ -61,10 +61,10 @@ resource "kubernetes_deployment" "stirling-pdf" {
           resources {
             requests = {
               cpu    = "25m"
-              memory = "1200Mi"
+              memory = "1536Mi"
             }
             limits = {
-              memory = "1200Mi"
+              memory = "1536Mi"
             }
           }
 

stacks/whisper/main.tf

@@ -38,7 +38,7 @@ resource "kubernetes_deployment" "whisper" {
     }
   }
   spec {
-    replicas = 1
+    replicas = 0 # Scaled down - GPU node memory pressure
     strategy {
       type = "Recreate"
     }
@@ -156,7 +156,7 @@ resource "kubernetes_deployment" "piper" {
     }
   }
   spec {
-    replicas = 1
+    replicas = 0 # Scaled down - GPU node memory pressure
     strategy {
       type = "Recreate"
     }