Merge branch 'main' of https://github.com/mitchross/talos-argocd-proxmox

Author: mitchross

infrastructure/controllers/argocd/kustomization.yaml

@@ -12,7 +12,7 @@ resources:
 helmCharts:
   - name: argo-cd
     repo: https://argoproj.github.io/argo-helm
-    version: "9.4.1" # It's good practice to pin the chart version
+    version: "9.4.4" # It's good practice to pin the chart version
     releaseName: argocd
     namespace: argocd
     valuesFile: values.yaml

infrastructure/controllers/reloader/kustomization.yaml

@@ -6,6 +6,6 @@ resources:
 helmCharts:
   - name: reloader
     repo: https://stakater.github.io/stakater-charts
-    version: 2.2.7
+    version: 2.2.8
     releaseName: reloader
     valuesFile: values.yaml

infrastructure/storage/snapshot-controller/values.yaml

@@ -5,7 +5,7 @@ replicaCount: 2
 
 image:
   repository: registry.k8s.io/sig-storage/snapshot-controller
-  tag: v8.4.0
+  tag: v8.5.0
 
 affinity:
   podAntiAffinity:

monitoring/prometheus-stack/kustomization.yaml

@@ -21,7 +21,7 @@ patchesStrategicMerge:
 helmCharts:
   - name: kube-prometheus-stack
     repo: https://prometheus-community.github.io/helm-charts
-    version: 81.6.2
+    version: 81.6.9
     releaseName: kube-prometheus-stack
     namespace: prometheus-stack
     valuesFile: values.yaml

my-apps/ai/comfyui/deployment.yaml

@@ -113,7 +113,7 @@ spec:
       - name: comfyui
         # Image from https://github.com/YanWenKun/ComfyUI-Docker/tree/main/cu128-megapak-pt29
         # renovate: datasource=docker depName=yanwk/comfyui-boot
-        image: yanwk/comfyui-boot:cu128-megapak-pt29@sha256:52c8939b54000970fa852973856a63e41db87aaecfd8006bb2d7326794330597
+        image: yanwk/comfyui-boot:cu128-megapak-pt29@sha256:778d8bd9d8e5ccf7ae4b8a6841f2ec484616d51c0c55ff473dffcbe817da3aee
         imagePullPolicy: IfNotPresent
         workingDir: /root
         ports:

my-apps/ai/llama-cpp/configmap.yaml

@@ -18,8 +18,6 @@ data:
     ctx-size = 65536
     n-gpu-layers = 99
     tensor-split = 1,1
-    cache-type-k = q8_0
-    cache-type-v = q4_0
     temp = 1.0
     top-p = 0.95
     top-k = 40
@@ -39,8 +37,6 @@ data:
     ctx-size = 32768
     n-gpu-layers = 99
     tensor-split = 1,1
-    cache-type-k = q8_0
-    cache-type-v = q4_0
     temp = 0.7
     top-p = 0.95
     top-k = 20
@@ -54,18 +50,16 @@ data:
     # 397B total / 17B active (MoE) - Unsloth Dynamic Q4_K_XL
     # WARNING: ~5-15 tok/s due to cpu-moe offloading. Quality over speed.
     # Natively multimodal (vision + language), 256K context native
-    # cpu-moe keeps attention on GPU, experts on CPU - MUCH faster than
+    # override-tensor keeps attention on GPU, experts on CPU - MUCH faster than
     # unified memory swapping (targeted offload vs indiscriminate CUDA paging)
     model = /models/UD-Q4_K_XL/Qwen3.5-397B-A17B-UD-Q4_K_XL-00001-of-00006.gguf
     alias = qwen3.5, qwen 3.5, general, experimental slow
     ctx-size = 32768
     n-gpu-layers = 99
     tensor-split = 1,1
-    cache-type-k = q8_0
-    cache-type-v = q4_0
-    cpu-moe = 1
-    temp = 0.6
+    override-tensor = exps=CPU
+    temp = 0.7
     top-p = 0.95
-    top-k = 20
+    top-k = 40
     min-p = 0.0
     jinja = 1

my-apps/ai/llama-cpp/deployment.yaml

@@ -48,6 +48,8 @@ spec:
             - "-fa"
             - "on"        # Explicitly set to 'on' so --jinja is read correctly
             - "--jinja"
+            - "--fit"           # Auto-fit dense layers to available VRAM
+            - "on"
             - "--no-mmap"       # Prevent page fault stalls - we have 400GB RAM to spare
             - "-b"
             - "4096"            # Larger logical batch for faster prompt processing
@@ -67,7 +69,7 @@ spec:
             - name: NVIDIA_DRIVER_CAPABILITIES
               value: "compute,utility"
             - name: GGML_CUDA_ENABLE_UNIFIED_MEMORY
-              value: "1" # Vital for Kimi-K2 1T model to bridge VRAM and 400GB RAM
+              value: "1" # Bridges VRAM and 400GB RAM for Qwen3.5-397B MoE expert offloading
             - name: GGML_CUDA_PEER_MAX_BATCH_SIZE
               value: "128"
             - name: CUDA_SCALE_LAUNCH_QUEUES

my-apps/ai/ollama/deployment.yaml

@@ -34,7 +34,7 @@ spec:
         effect: "NoSchedule"
       containers:
       - name: ollama
-        image: ollama/ollama:0.15.6
+        image: ollama/ollama:0.17.0
         imagePullPolicy: IfNotPresent
         ports:
         - containerPort: 11434

my-apps/ai/open-webui/deployment.yaml

@@ -28,7 +28,7 @@ spec:
       containers:
       - name: open-webui
         # renovate: datasource=docker depName=ghcr.io/open-webui/open-webui
-        image: ghcr.io/open-webui/open-webui:cuda@sha256:268d2b1b1f66f2d3f9a4ac3e0f06e0f545e9d6aaf71027115861874120e7edde
+        image: ghcr.io/open-webui/open-webui:cuda@sha256:b5e6c753c8c43cdcb987910fbe257c31abef134d44bc84683775e6a1e4f3fafe
         imagePullPolicy: IfNotPresent
         ports:
         - containerPort: 8080

my-apps/ai/open-webui/function-loader-job.yaml

@@ -21,7 +21,7 @@ spec:
         effect: "NoSchedule"
       containers:
       - name: loader
-        image: ghcr.io/open-webui/open-webui:cuda@sha256:268d2b1b1f66f2d3f9a4ac3e0f06e0f545e9d6aaf71027115861874120e7edde
+        image: ghcr.io/open-webui/open-webui:cuda@sha256:b5e6c753c8c43cdcb987910fbe257c31abef134d44bc84683775e6a1e4f3fafe
         command:
         - python3
         - -c