add ai workloads

kubernetes/apps/ai/kustomization.yaml

@@ -0,0 +1,11 @@
+---
+# yaml-language-server: $schema=https://json.schemastore.org/kustomization
+apiVersion: kustomize.config.k8s.io/v1beta1
+kind: Kustomization
+resources:
+  # Pre Flux-Kustomizations
+  - ./namespace.yaml
+  # Flux-Kustomizations
+  - ./ollama/ks.yaml
+  - ./open-webui/ks.yaml
+  - ./stable-diffusion/ks.yaml

kubernetes/apps/ai/namespace.yaml

@@ -0,0 +1,8 @@
+---
+apiVersion: v1
+kind: Namespace
+metadata:
+  name: ai
+  labels:
+    kustomize.toolkit.fluxcd.io/prune: disabled
+    volsync.backube/privileged-movers: "true"

kubernetes/apps/ai/ollama/app/helmrelease.yaml

@@ -0,0 +1,88 @@
+---
+# yaml-language-server: $schema=https://raw.githubusercontent.com/bjw-s/helm-charts/main/charts/other/app-template/schemas/helmrelease-helm-v2beta2.schema.json
+apiVersion: helm.toolkit.fluxcd.io/v2
+kind: HelmRelease
+metadata:
+  name: &app ollama
+spec:
+  interval: 30m
+  chart:
+    spec:
+      chart: app-template
+      version: 3.4.0
+      sourceRef:
+        kind: HelmRepository
+        name: bjw-s
+        namespace: flux-system
+  install:
+    remediation:
+      retries: 3
+  upgrade:
+    cleanupOnFail: true
+    remediation:
+      retries: 3
+      strategy: rollback
+  values:
+    controllers:
+      ollama:
+        annotations:
+          reloader.stakater.com/auto: "true"
+        pod:
+          nodeSelector:
+            nvidia.com/gpu.present: "true"
+          runtimeClassName: nvidia
+        containers:
+          app:
+            image:
+              repository: docker.io/ollama/ollama
+              tag: 0.3.8
+            env:
+              - name: OLLAMA_HOST
+                value: 0.0.0.0
+              - name: OLLAMA_ORIGINS
+                value: "*"
+              - name: OLLAMA_MODELS
+                value: &modelPath "/models"
+              - name: OLLAMA_KEEP_ALIVE
+                value: "24h"
+            resources:
+              requests:
+                nvidia.com/gpu: 1 # requesting 1 GPU
+                cpu: 500m
+                memory: 2Gi
+              limits:
+                memory: 16Gi
+                nvidia.com/gpu: 1 # requesting 1 GPU
+    service:
+      app:
+        controller: ollama
+        ports:
+          http:
+            port: 11434
+    ingress:
+      app:
+        enabled: true
+        className: internal-nginx
+        hosts:
+          - host: &host "{{ .Release.Name }}.jahanson.tech"
+            paths:
+              - path: /
+                service:
+                  identifier: app
+                  port: http
+        tls:
+          - hosts:
+              - *host
+    persistence:
+      models:
+        enabled: true
+        existingClaim: ollama-models
+        advancedMounts:
+          ollama:
+            app:
+              - path: *modelPath
+      config:
+        enabled: true
+        existingClaim: ollama
+        globalMounts:
+          - path: /root/.ollama

kubernetes/apps/ai/ollama/app/kustomization.yaml

@@ -0,0 +1,8 @@
+---
+# yaml-language-server: $schema=https://json.schemastore.org/kustomization
+apiVersion: kustomize.config.k8s.io/v1beta1
+kind: Kustomization
+resources:
+  - ./helmrelease.yaml
+  - ./pvc.yaml
+  - ../../../../templates/volsync

kubernetes/apps/ai/ollama/app/pvc.yaml

@@ -0,0 +1,12 @@
+---
+apiVersion: v1
+kind: PersistentVolumeClaim
+metadata:
+  name: ollama-models
+spec:
+  accessModes:
+    - ReadWriteOnce
+  resources:
+    requests:
+      storage: 100Gi
+  storageClassName: openebs-hostpath

kubernetes/apps/ai/ollama/ks.yaml

@@ -0,0 +1,32 @@
+---
+# yaml-language-server: $schema=https://ks.hsn.dev/kustomize.toolkit.fluxcd.io/kustomization_v1.json
+apiVersion: kustomize.toolkit.fluxcd.io/v1
+kind: Kustomization
+metadata:
+  name: &app ollama
+  namespace: flux-system
+spec:
+  targetNamespace: ai
+  commonMetadata:
+    labels:
+      app.kubernetes.io/name: *app
+  dependsOn:
+    - name: nvidia-device-plugin
+    - name: node-feature-discovery
+    - name: volsync
+    - name: openebs
+  path: ./kubernetes/apps/ai/ollama/app
+  prune: true
+  sourceRef:
+    kind: GitRepository
+    name: theshire
+  wait: false
+  interval: 30m
+  retryInterval: 1m
+  timeout: 5m
+  postBuild:
+    substitute:
+      APP: *app
+      VOLSYNC_CAPACITY: 1Gi
+      VOLSYNC_STORAGECLASS: openebs-zfs
+      VOLSYNC_SNAPSHOTCLASS: openebs-zfs

kubernetes/apps/ai/open-webui/app/helmrelease.yaml

@@ -0,0 +1,77 @@
+---
+# yaml-language-server: $schema=https://raw.githubusercontent.com/bjw-s/helm-charts/main/charts/other/app-template/schemas/helmrelease-helm-v2beta2.schema.json
+apiVersion: helm.toolkit.fluxcd.io/v2
+kind: HelmRelease
+metadata:
+  name: &app open-webui
+spec:
+  interval: 30m
+  chart:
+    spec:
+      chart: app-template
+      version: 3.4.0
+      sourceRef:
+        kind: HelmRepository
+        name: bjw-s
+        namespace: flux-system
+  dependsOn:
+    - name: ollama
+  install:
+    remediation:
+      retries: 3
+  upgrade:
+    cleanupOnFail: true
+    remediation:
+      retries: 3
+      strategy: rollback
+  values:
+    controllers:
+      open-webui:
+        annotations:
+          reloader.stakater.com/auto: "true"
+        containers:
+          app:
+            image:
+              repository: ghcr.io/open-webui/open-webui
+              tag: v0.3.16
+            env:
+              - name: OLLAMA_BASE_URL
+                value: http://ollama.ai.svc.cluster.local:11434
+              - name: ENABLE_RAG_WEB_SEARCH
+                value: true
+              - name: RAG_WEB_SEARCH_ENGINE
+                value: searxng
+              - name: SEARXNG_QUERY_URL
+                value: http://searxng.default.svc.cluster.local:8080/search?q=<query>
+            resources:
+              requests:
+                cpu: 500m
+                memory: 2Gi
+              limits:
+                memory: 2Gi
+    service:
+      app:
+        controller: open-webui
+        ports:
+          http:
+            port: 8080
+    ingress:
+      app:
+        enabled: true
+        className: internal-nginx
+        hosts:
+          - host: &host "chat.jahanson.tech"
+            paths:
+              - path: /
+                service:
+                  identifier: app
+                  port: http
+        tls:
+          - hosts:
+              - *host
+    persistence:
+      config:
+        enabled: true
+        existingClaim: *app
+        globalMounts:
+          - path: /app/backend/data

kubernetes/apps/ai/open-webui/app/kustomization.yaml

@@ -0,0 +1,8 @@
+---
+# yaml-language-server: $schema=https://json.schemastore.org/kustomization
+apiVersion: kustomize.config.k8s.io/v1beta1
+kind: Kustomization
+resources:
+  - ./helmrelease.yaml
+  - ../../../../templates/volsync
+  - ../../../../templates/gatus/internal

kubernetes/apps/ai/open-webui/ks.yaml

@@ -0,0 +1,29 @@
+---
+# yaml-language-server: $schema=https://ks.hsn.dev/kustomize.toolkit.fluxcd.io/kustomization_v1.json
+apiVersion: kustomize.toolkit.fluxcd.io/v1
+kind: Kustomization
+metadata:
+  name: &app open-webui
+  namespace: flux-system
+spec:
+  targetNamespace: ai
+  commonMetadata:
+    labels:
+      app.kubernetes.io/name: *app
+  dependsOn:
+    - name: volsync
+    - name: ollama
+  path: ./kubernetes/apps/ai/open-webui/app
+  prune: true
+  sourceRef:
+    kind: GitRepository
+    name: theshire
+  wait: false
+  interval: 30m
+  retryInterval: 1m
+  timeout: 5m
+  postBuild:
+    substitute:
+      APP: *app
+      VOLSYNC_CAPACITY: 5Gi
+      GATUS_SUBDOMAIN: chat

kubernetes/apps/ai/stable-diffusion/comfyui/helmrelease.yaml

@@ -0,0 +1,82 @@
+---
+# yaml-language-server: $schema=https://raw.githubusercontent.com/bjw-s/helm-charts/main/charts/other/app-template/schemas/helmrelease-helm-v2beta2.schema.json
+apiVersion: helm.toolkit.fluxcd.io/v2
+kind: HelmRelease
+metadata:
+  name: &app comfyui
+spec:
+  interval: 30m
+  chart:
+    spec:
+      chart: app-template
+      version: 3.4.0
+      sourceRef:
+        kind: HelmRepository
+        name: bjw-s
+        namespace: flux-system
+  install:
+    remediation:
+      retries: 3
+  upgrade:
+    cleanupOnFail: true
+    remediation:
+      retries: 3
+      strategy: rollback
+  values:
+    controllers:
+      comfyui:
+        annotations:
+          reloader.stakater.com/auto: "true"
+        pod:
+          nodeSelector:
+            nvidia.com/gpu.present: "true"
+          runtimeClassName: nvidia
+        containers:
+          app:
+            image:
+              repository: docker.io/jahanson/comfyui
+              tag: v0.0.1
+            resources:
+              requests:
+                nvidia.com/gpu: 1 # requesting 1 GPU
+                cpu: 500m
+                memory: 2Gi
+              limits:
+                memory: 60Gi
+                nvidia.com/gpu: 1 # requesting 1 GPU
+    service:
+      app:
+        controller: comfyui
+        ports:
+          http:
+            port: 7860
+    ingress:
+      app:
+        enabled: true
+        className: internal-nginx
+        hosts:
+          - host: &host "{{ .Release.Name }}.jahanson.tech"
+            paths:
+              - path: /
+                service:
+                  identifier: app
+                  port: http
+        tls:
+          - hosts:
+              - *host
+    persistence:
+      models:
+        enabled: true
+        existingClaim: stablediffusion-checkpoints
+        globalMounts:
+          - path: /data/models
+      config:
+        enabled: true
+        existingClaim: comfyui
+        globalMounts:
+          - path: /data/config
+      output:
+        enabled: true
+        type: emptyDir
+        globalMounts:
+          - path: /output

kubernetes/apps/ai/stable-diffusion/comfyui/kustomization.yaml

@@ -0,0 +1,9 @@
+---
+# yaml-language-server: $schema=https://json.schemastore.org/kustomization
+apiVersion: kustomize.config.k8s.io/v1beta1
+kind: Kustomization
+resources:
+  - ./helmrelease.yaml
+  - ./pvc.yaml
+  - ../../../../templates/volsync
+  - ../../../../templates/gatus/internal

kubernetes/apps/ai/stable-diffusion/comfyui/pvc.yaml

@@ -0,0 +1,12 @@
+---
+apiVersion: v1
+kind: PersistentVolumeClaim
+metadata:
+  name: stablediffusion-checkpoints
+spec:
+  accessModes:
+    - ReadWriteOnce
+  resources:
+    requests:
+      storage: 300Gi
+  storageClassName: openebs-hostpath

kubernetes/apps/ai/stable-diffusion/ks.yaml

@@ -0,0 +1,31 @@
+---
+# yaml-language-server: $schema=https://ks.hsn.dev/kustomize.toolkit.fluxcd.io/kustomization_v1.json
+apiVersion: kustomize.toolkit.fluxcd.io/v1
+kind: Kustomization
+metadata:
+  name: &app comfyui
+  namespace: flux-system
+spec:
+  targetNamespace: ai
+  commonMetadata:
+    labels:
+      app.kubernetes.io/name: *app
+  dependsOn:
+    - name: nvidia-device-plugin
+    - name: node-feature-discovery
+    - name: volsync
+    - name: rook-ceph-cluster
+  path: ./kubernetes/apps/ai/stable-diffusion/comfyui
+  prune: true
+  sourceRef:
+    kind: GitRepository
+    name: theshire
+  wait: false
+  interval: 30m
+  retryInterval: 1m
+  timeout: 5m
+  postBuild:
+    substitute:
+      APP: *app
+      VOLSYNC_CAPACITY: 5Gi
+      GATUS_SUBDOMAIN: comfyui