Update ISO installer tiers and backend deployment

ansible/roles/ollama/tasks/main.yml

@@ -28,7 +28,7 @@
       Environment="PATH=/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin"
       Environment="OLLAMA_HOST=0.0.0.0:11434"
       Environment="OLLAMA_MODELS=/opt/cezen/models/ollama"
-      Environment="CUDA_VISIBLE_DEVICES=0,1,2"
+      Environment="CUDA_VISIBLE_DEVICES=0"
 
       [Install]
       WantedBy=multi-user.target
@@ -54,50 +54,49 @@
     port: 11434
     timeout: 60
 
-- name: Pull default models (Llama 3.1 8B + Mistral 7B)
+- name: Select tier model set
+  set_fact:
+    ollama_models: >-
+      {{
+        {
+          'starter': ['phi3:mini', 'nomic-embed-text'],
+          'basic': ['llama3.1:8b', 'mistral:7b', 'nomic-embed-text', 'codellama:13b'],
+          'entry': ['llama3.1:8b', 'mistral:7b', 'nomic-embed-text', 'codellama:13b'],
+          'pro': ['llama3.1:8b', 'mistral:7b', 'nomic-embed-text', 'codellama:13b', 'llama3.1:70b', 'mixtral:8x7b', 'deepseek-coder-v2:16b'],
+          'max': ['llama3.1:8b', 'mistral:7b', 'nomic-embed-text', 'codellama:13b', 'llama3.1:70b', 'mixtral:8x7b', 'deepseek-coder-v2:16b', 'llama3.1:405b', 'mixtral:8x22b']
+        }.get(tier | default('basic'), ['llama3.1:8b', 'mistral:7b', 'nomic-embed-text'])
+      }}
+
+- name: Pull tier Ollama models
   become_user: cezen
   command: ollama pull {{ item }}
-  loop:
-    - llama3.1:8b
-    - mistral:7b
+  loop: "{{ ollama_models }}"
   environment:
     OLLAMA_HOST: "http://localhost:11434"
   retries: 3
   delay: 15
-  # NOTE: Models are large (~5GB each). This step takes time on first run.
-  # Skip by setting: ansible-playbook ... -e "skip_model_pull=true"
+  # NOTE: Pro/Max models are very large. Skip with --skip-model-pull for
+  # bandwidth-constrained installs, then run models/pull-models.sh later.
   when: not (skip_model_pull | default(false))
 
 # Open WebUI (chat interface on top of Ollama)
-- name: Deploy Open WebUI via Docker
-  community.docker.docker_container:
-    name: open-webui
-    image: ghcr.io/open-webui/open-webui:main
-    state: started
-    restart_policy: always
-    ports:
-      - "3001:8080"
-    volumes:
-      - open-webui:/app/backend/data
-    env:
-      OLLAMA_BASE_URL: "http://host-gateway:11434"
-    etc_hosts:
-      host-gateway: "172.17.0.1"
-  # Note: Requires docker community collection. Install with:
-  # ansible-galaxy collection install community.docker
-  ignore_errors: true  # Falls back gracefully if docker collection not available
-
-- name: Alternative Open WebUI start (if community.docker not available)
+- name: Start Open WebUI via Docker CLI
   shell: |
-    docker run -d \
-      --name open-webui \
-      --restart always \
-      -p 3001:8080 \
-      --add-host=host-gateway:172.17.0.1 \
-      -v open-webui:/app/backend/data \
-      -e OLLAMA_BASE_URL=http://host-gateway:11434 \
-      ghcr.io/open-webui/open-webui:main
+    if docker ps -a --format '{{ "{{" }}.Names{{ "}}" }}' | grep -qx open-webui; then
+      docker start open-webui
+    else
+      docker run -d \
+        --name open-webui \
+        --restart always \
+        -p 3001:8080 \
+        --add-host=host-gateway:172.17.0.1 \
+        -v open-webui:/app/backend/data \
+        -e OLLAMA_BASE_URL=http://host-gateway:11434 \
+        ghcr.io/open-webui/open-webui:main
+    fi
   args:
     executable: /bin/bash
   register: webui_result
-  failed_when: webui_result.rc != 0 and 'already in use' not in webui_result.stderr
+  changed_when: webui_result.rc == 0
+  failed_when: webui_result.rc != 0 and 'already in use' not in (webui_result.stderr | default(''))
+  ignore_errors: true