Update Marketplace output for Rag and Jupyter Application (GoogleClou…

…dPlatform#324)

* Add outputs for Jupyter, Ray & Rag Marketplace

applications/jupyter/metadata.display.yaml

@@ -164,11 +164,20 @@ spec:
     runtime:
       outputMessage: Deployment can take several minutes to complete.
       suggestedActions:
-        - heading: No external IP configured
-          description: The application has been configured with no external IP or domain.
+        - heading: "Access JupyterHub Application"
+          description: |-
+            Go to the JupyterHub Application, log in with <i>Jupyterhub User</i> and <i>Jupyterhub Password</i> (from the Outputs section).
+            Once logged in, choose the appropriate preset and execute notebooks. 
+          showIf: '!variables.add_auth'
+        - heading: "Access JupyterHub Application"
+          description: |-
+            Go to the JupyterHub Application, log in with your organization's credentials. Once logged in, choose the appropriate preset and execute notebooks. 
+            Note: Application endpoint can take around 15-20 minutes to become accessible.
+          showIf: 'variables.add_auth'
       outputs:
         jupyterhub_password: {}
+        jupyterhub_user: {}
         jupyterhub_uri:
           openInNewTab: true
           showInNotification: true
-          label: Go to Jupyter Application
+          label: Go to JupyterHub Application

applications/jupyter/metadata.yaml

@@ -132,4 +132,8 @@ spec:
         defaultValue: jupyter-service-account
     outputs:
       - name: jupyterhub_password
+        description: "JupyterHub password is only required for standard authentication. Ignore in case of IAP authentication"
+      - name: jupyterhub_user
+        description: "JupyterHub user is only required for standard authentication. Ignore in case of IAP authentication"
       - name: jupyterhub_uri
+        description: "JupyterHub Endpoint to access user interface. In case of private IP consider port-forwarding."

applications/jupyter/outputs.tf

@@ -16,6 +16,10 @@ output "jupyterhub_uri" {
   value = "http://${module.jupyterhub.jupyterhub_uri}"
 }
 
+output "jupyterhub_user" {
+  value = module.jupyterhub.jupyterhub_user
+}
+
 output "jupyterhub_password" {
   value     = module.jupyterhub.jupyterhub_password
   sensitive = true

applications/rag/metadata.display.yaml

@@ -186,19 +186,34 @@ spec:
     runtime:
       outputMessage: Deployment can take several minutes to complete.
       suggestedActions:
-        - heading: No external IP configured
+        - heading: "Step 1: Generate Vector Embeddings for the Dataset"
           description: |-
-            The VM instance has been configured with no external IP or internet access. Connecting 
-            to it directly may not be possible. Consider using a <a href="https://cloud.google.com/compute/docs/instances/connecting-advanced#bastion_host">bastion host</a> 
-            to access this VM.
+            Go to the JupyterHub Application, log in with <i>Jupyterhub User</i> and <i>Jupyterhub Password</i> (from the Outputs section).
+            Once logged in, choose the <b>CPU</b> preset. Go to File -> Open From URL & upload and execute the notebook 
+            <a href="https://raw.githubusercontent.com/GoogleCloudPlatform/ai-on-gke/main/applications/rag/example_notebooks/rag-kaggle-ray-sql-latest.ipynb"><i>rag-kaggle-ray-sql.ipynb</i></a>.
+            Follow the <a href="https://github.com/GoogleCloudPlatform/ai-on-gke/blob/main/applications/rag/README.md#vector-embeddings-for-dataset">README.md</a> for detailed instructions.
+          showIf: '!variables.add_auth'
+        - heading: "Step 1: Generate Vector Embeddings for the Dataset"
+          description: |-
+            Go to the JupyterHub Application, log in with your organization's credentials.
+            Once logged in, choose the <b>CPU</b> preset. Go to File -> Open From URL & upload and execute the notebook 
+            <a href="https://raw.githubusercontent.com/GoogleCloudPlatform/ai-on-gke/main/applications/rag/example_notebooks/rag-kaggle-ray-sql-latest.ipynb"><i>rag-kaggle-ray-sql.ipynb</i></a>.
+            Follow the <a href="https://github.com/GoogleCloudPlatform/ai-on-gke/blob/main/applications/rag/README.md#vector-embeddings-for-dataset">README.md</a> for detailed instructions.
+          showIf: 'variables.add_auth'
+        - heading: "Step 2: Prompt the Inference Server via a Chatbot"
+          description: |-
+            Go to the RAG Frontend Application, scroll to <i>Ports</i> section and initiate <b>PORT FORWARDING</b> (Run in Cloud Shell) to the front end application.
+            Launch Fronted Chat app via <b>OPEN IN WEB PREVIEW</b> button and prompt the LLM. This will fetch context related to your prompt from the generated vector embeddings,
+            augment the original prompt with the context & query the inference model (mistral-7b) with the augmented prompt.
       outputs:
         jupyterhub_password: {}
+        jupyterhub_user: {}
         jupyterhub_uri:
           openInNewTab: true
           showInNotification: true
-          label: Go to Jupyter Application
+          label: Go to JupyterHub Application
         frontend_uri:
           openInNewTab: true
           showInNotification: true
-          label: Go to Rag Frontend Application
+          label: Go to RAG Frontend Application
 

applications/rag/metadata.yaml

@@ -134,5 +134,10 @@ spec:
         defaultValue: "pgvector-instance"
     outputs:
       - name: jupyterhub_password
-      - name: jupyterhub_uri
+        description: "JupyterHub password is only required for standard authentication. Ignore, in case of IAP authentication"
+      - name: jupyterhub_user
+        description: "JupyterHub user is only required for standard authentication. Ignore, in case of IAP authentication"
       - name: frontend_uri
+        description: "RAG Frontend Endpoint to access user interface. In case of private IP, consider port-forwarding."
+      - name: jupyterhub_uri
+        description: "JupyterHub Endpoint to access user interface. In case of private IP, consider port-forwarding."

applications/rag/outputs.tf

@@ -13,14 +13,22 @@
 # limitations under the License.
 
 output "jupyterhub_uri" {
-  value = "http://${module.jupyterhub.jupyterhub_uri}"
+  description = "JupyterHub Endpoint to access user interface. In case of private IP, consider port-forwarding."
+  value       = "http://${module.jupyterhub.jupyterhub_uri}"
+}
+
+output "jupyterhub_user" {
+  description = "JupyterHub user is only required for standard authentication. Ignore, in case of IAP authentication"
+  value       = module.jupyterhub.jupyterhub_user
 }
 
 output "jupyter_password" {
-  value     = module.jupyterhub.jupyterhub_password
-  sensitive = true
+  description = "JupyterHub password is only required for standard authentication. Ignore, in case of IAP authentication"
+  value       = module.jupyterhub.jupyterhub_password
+  sensitive   = true
 }
 
 output "frontend_uri" {
-  value = module.frontend.frontend_uri != "" ? "http://${module.frontend.frontend_uri}" : local.frontend_default_uri
+  description = "RAG Frontend Endpoint to access user interface. In case of private IP, consider port-forwarding."
+  value       = module.frontend.frontend_uri != "" ? "http://${module.frontend.frontend_uri}" : local.frontend_default_uri
 }

applications/ray/main.tf

@@ -62,6 +62,7 @@ locals {
 
 locals {
   workload_identity_service_account = var.goog_cm_deployment_name != "" ? "${var.goog_cm_deployment_name}-${var.workload_identity_service_account}" : var.workload_identity_service_account
+  ray_cluster_default_uri           = "https://console.cloud.google.com/kubernetes/service/${var.cluster_location}/${var.cluster_name}/${var.kubernetes_namespace}/${var.ray_cluster_name}-kuberay-head-svc/overview?project=${var.project_id}"
 }
 
 provider "kubernetes" {
@@ -144,6 +145,7 @@ module "kuberay-cluster" {
   count                  = var.create_ray_cluster == true ? 1 : 0
   source                 = "../../modules/kuberay-cluster"
   providers              = { helm = helm.ray, kubernetes = kubernetes.ray }
+  name                   = var.ray_cluster_name
   namespace              = var.kubernetes_namespace
   project_id             = var.project_id
   enable_tpu             = local.enable_tpu

applications/ray/metadata.display.yaml

@@ -115,4 +115,18 @@ spec:
         - name: ray
           title: Ray Application
         - name: iap_auth
-          title: IAP Authentication
+          title: IAP Authentication
+    runtime:
+      outputMessage: Deployment can take several minutes to complete.
+      suggestedActions:
+        - heading: "Connect to Ray Cluster"
+          description: |-
+            Connect to Ray Cluster, scroll to <i>Ports</i> section and initiate <b>PORT FORWARDING</b> (Run in Cloud Shell) to the ray cluster head node.
+        - heading: "Submit Ray Jobs"
+          description: |-
+            Open the ray dashboard via the <b>OPEN IN WEB PREVIEW</b> button and submit ray jobs following these <a href="https://github.com/GoogleCloudPlatform/ai-on-gke/tree/main/applications/ray#using-ray-with-ray-jobs-api">instructions</a>. 
+      outputs:
+        ray_cluster_uri:
+          openInNewTab: true
+          showInNotification: true
+          label: Connect to Ray Cluster

applications/ray/metadata.yaml

@@ -80,3 +80,4 @@ spec:
         defaultValue: v2.7.1
     outputs:
       - name: grafana_uri
+      - name: ray_cluster_uri

applications/ray/outputs.tf

@@ -1,3 +1,7 @@
 output "grafana_uri" {
   value = module.kuberay-monitoring[0].grafana_uri
 }
+
+output "ray_cluster_uri" {
+  value = module.kuberay-cluster[0].ray_cluster_uri != "" ? "http://${module.kuberay-cluster[0].ray_cluster_uri}" : local.ray_cluster_default_uri
+}

applications/ray/variables.tf

@@ -75,6 +75,11 @@ variable "create_ray_cluster" {
   default = false
 }
 
+variable "ray_cluster_name" {
+  type    = string
+  default = "example-cluster"
+}
+
 variable "enable_gpu" {
   type    = bool
   default = false

applications/ray/workloads.tfvars

@@ -27,7 +27,7 @@ cluster_location = "us-central1"
 #######################################################
 
 ## GKE environment variables
-ray_namespace = "ml"
+kubernetes_namespace = "ml"
 
 # Creates a google service account & k8s service account & configures workload identity with appropriate permissions.
 # Set to false & update the variable `workload_identity_service_account` to use an existing IAM service account.
@@ -38,4 +38,5 @@ workload_identity_service_account = "ray-service-account"
 create_gcs_bucket               = true
 gcs_bucket                      = "ray-bucket-zydg"
 create_ray_cluster              = true
+ray_cluster_name                = "example-cluster"
 enable_grafana_on_ray_dashboard = false

modules/jupyter/outputs.tf

@@ -16,6 +16,10 @@ output "jupyterhub_uri" {
   value = var.add_auth ? module.iap_auth[0].domain : (data.kubernetes_service.jupyter-ingress.status != null ? (data.kubernetes_service.jupyter-ingress.status[0].load_balancer != null ? "${data.kubernetes_service.jupyter-ingress.status[0].load_balancer[0].ingress[0].ip}" : "") : "")
 }
 
+output "jupyterhub_user" {
+  value = var.add_auth ? "" : "admin"
+}
+
 output "jupyterhub_password" {
   value     = var.add_auth ? "" : random_password.generated_password[0].result
   sensitive = true

modules/kuberay-cluster/kuberay.tf → modules/kuberay-cluster/main.tf

@@ -23,7 +23,7 @@ locals {
 }
 
 resource "helm_release" "ray-cluster" {
-  name             = "example-cluster"
+  name             = var.name
   repository       = "https://ray-project.github.io/kuberay-helm/"
   chart            = "ray-cluster"
   namespace        = var.namespace
@@ -69,7 +69,8 @@ resource "helm_release" "ray-cluster" {
 resource "kubernetes_service" "tpu-worker-svc" {
   count = var.enable_tpu ? 1 : 0
   metadata {
-    name = "${helm_release.ray-cluster.name}-kuberay-tpu-worker-svc"
+    name      = "${helm_release.ray-cluster.name}-kuberay-tpu-worker-svc"
+    namespace = var.namespace
   }
   spec {
     selector = {
@@ -78,3 +79,11 @@ resource "kubernetes_service" "tpu-worker-svc" {
     cluster_ip = "None"
   }
 }
+
+data "kubernetes_service" "head-svc" {
+  metadata {
+    name      = "${helm_release.ray-cluster.name}-kuberay-head-svc"
+    namespace = var.namespace
+  }
+  depends_on = [helm_release.ray-cluster]
+}

modules/kuberay-cluster/outputs.tf

@@ -0,0 +1,20 @@
+# Copyright 2023 Google LLC
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+
+output "ray_cluster_uri" {
+  value = data.kubernetes_service.head-svc.status != null ? (
+    length(data.kubernetes_service.head-svc.status[0].load_balancer) > 0 ? (
+      length(data.kubernetes_service.head-svc.status[0].load_balancer[0].ingress) > 0 ?
+  "${data.kubernetes_service.head-svc.status[0].load_balancer[0].ingress[0].ip}" : "") : "") : ""
+}

modules/kuberay-cluster/variables.tf

@@ -17,6 +17,12 @@ variable "project_id" {
   description = "GCP project id"
 }
 
+variable "name" {
+  type        = string
+  description = "Name of the ray cluster"
+  default     = "example-cluster"
+}
+
 variable "db_region" {
   type        = string
   description = "Cloud SQL instance region"