Frontend

stacks/ai/ai-frontend.nomad

@@ -3,51 +3,100 @@ job "ai-frontend" {
   region      = "home"
   type        = "service"
 
+  # --- OpenWebUI ---
   group "openwebui" {
     count = 1
+
+    # PIN TO HP1 (Change "hp-mini-1" to your actual node name if different)
+    constraint {
+      attribute = "${attr.unique.hostname}"
+      value     = "hp-mini-1" 
+    }
+
     network {
       port "http" { to = 8080 }
     }
+
     service {
       name = "openwebui"
       port = "http"
       tags = ["traefik.enable=true"]
-    }
+      
-    task "server" {
+      check {
-      driver = "podman"
+        type     = "http"
-      env {
+        path     = "/health"
-        OLLAMA_BASE_URL = "http://ollama.service.consul:11434"
+        interval = "20s"
-      }
+        timeout  = "2s"
-      config {
-        image = "ghcr.io/open-webui/open-webui:main"
-        ports = ["http"]
-        volumes = ["/opt/homelab/openwebui:/app/backend/data"]
-      }
-      resources { cpu = 1000; memory = 1024 }
       }
     }
 
+    task "server" {
+      driver = "podman"
+      
+      env {
+        # Magic: Uses Consul to find the P52 automatically
+        OLLAMA_BASE_URL = "http://ollama.service.consul:11434"
+      }
+
+      config {
+        image = "ghcr.io/open-webui/open-webui:main"
+        ports = ["http"]
+        volumes = [
+          "/mnt/local-ssd/nomad/stacks/ai/ai-frontend/openwebui:/app/backend/data"
+        ]
+      }
+      
+      resources {
+        cpu    = 1000
+        memory = 1024
+      }
+    }
+  }
+
+  # --- LobeChat ---
   group "lobechat" {
     count = 1
+    
+    # PIN TO HP1
+    constraint {
+      attribute = "${attr.unique.hostname}"
+      value     = "hp-mini-1"
+    }
+
     network {
       port "http" { to = 3210 }
     }
+
     service {
       name = "lobechat"
       port = "http"
       tags = ["traefik.enable=true"]
+      
+      check {
+        type     = "http"
+        path     = "/api/health"
+        interval = "20s"
+        timeout  = "2s"
       }
+    }
+
     task "server" {
       driver = "podman"
+      
       env {
         OLLAMA_PROXY_URL = "http://ollama.service.consul:11434"
         ACCESS_CODE      = "securepassword123"
       }
+
       config {
         image = "docker.io/lobehub/lobe-chat"
         ports = ["http"]
       }
-      resources { cpu = 500; memory = 512 }
+      
+      resources {
+        cpu    = 500
+        memory = 512
+      }
     }
   }
 }