Add MS SQL Server as a DB Provider for RAG backend

Makefile

@@ -1,3 +1,8 @@
+# Azure gpu vars
+GPU_VM_SIZE ?= Standard_NC8as_T4_v3
+GPU_REPLICAS ?= 1
+OVERRIDE_ZONE ?=
+
 .PHONY: default
 default: help
 
@@ -16,8 +21,13 @@ install: operator-deploy post-install ## installs the pattern and loads the secr
 	@echo "Installed"
 
 .PHONY: create-gpu-machineset
-create-gpu-machineset: ## Creates a gpu machineset
-	ansible-playbook ansible/playbooks/create-gpu-machine-set.yaml
+create-gpu-machineset: ## Creates a gpu machineset for AWS
+	ansible-playbook ansible/playbooks/create-gpu-machineset.yaml
+
+.PHONY: create-gpu-machineset-azure
+create-gpu-machineset-azure: ## Creates an Azure GPU machineset (overrides: GPU_VM_SIZE, GPU_REPLICAS, OVERRIDE_ZONE)
+	ansible-playbook ansible/playbooks/create-gpu-machineset-azure.yaml \
+		-e "gpu_vm_size=$(GPU_VM_SIZE) gpu_replicas=$(GPU_REPLICAS) override_zone=$(OVERRIDE_ZONE)"
 
 .PHONY: post-install
 post-install: ## Post-install tasks

ansible/playbooks/create-gpu-machineset-azure.yaml

@@ -0,0 +1,65 @@
+- name: Generate GPU MachineSet for Azure-based clusters
+  hosts: localhost
+  connection: local
+  gather_facts: false
+  vars:
+    gpu_vm_size: Standard_NC8as_T4_v3
+    gpu_replicas: 1
+    override_zone: ""
+    namespace: openshift-machine-api
+  tasks:
+    - name: Get cluster infrastructure object
+      kubernetes.core.k8s_info:
+        api_version: config.openshift.io/v1
+        kind: Infrastructure
+        name: cluster
+      register: infra_info
+
+    - name: Save cluster ID
+      set_fact:
+        cluster_id: "{{ infra_info.resources[0].status.infrastructureName }}"
+
+    - name: Gather all MachineSets
+      kubernetes.core.k8s_info:
+        api_version: machine.openshift.io/v1beta1
+        kind: MachineSet
+        namespace: "{{ namespace }}"
+      register: ms_list
+
+    - name: Pick the first *worker* MachineSet as a template
+      set_fact:
+        base_ms: "{{ item }}"
+      loop: "{{ ms_list.resources | sort(attribute='metadata.name') }}"
+      when:
+        - "'worker' in (item.metadata.labels['machine.openshift.io/cluster-api-machine-role'] | default(''))"
+      run_once: true
+
+    - name: Extract provider-specific details from the base MachineSet
+      set_fact:
+        azure_location: "{{ base_ms.spec.template.spec.providerSpec.value.location }}"
+        base_zone:      "{{ base_ms.spec.template.spec.providerSpec.value.zone }}"
+        resource_group: "{{ base_ms.spec.template.spec.providerSpec.value.resourceGroup }}"
+        network_resource_group: "{{ base_ms.spec.template.spec.providerSpec.value.networkResourceGroup }}"
+        vnet:           "{{ base_ms.spec.template.spec.providerSpec.value.vnet }}"
+        subnet:         "{{ base_ms.spec.template.spec.providerSpec.value.subnet }}"
+        image:          "{{ base_ms.spec.template.spec.providerSpec.value.image }}"
+        os_disk:        "{{ base_ms.spec.template.spec.providerSpec.value.osDisk }}"
+        user_data:      "{{ base_ms.spec.template.spec.providerSpec.value.userDataSecret }}"
+        cred_secret:    "{{ base_ms.spec.template.spec.providerSpec.value.credentialsSecret }}"
+        public_ip:      "{{ base_ms.spec.template.spec.providerSpec.value.publicIP | default(false) }}"
+
+    - name: Decide which availability zone to use
+      set_fact:
+        gpu_zone: "{{ (override_zone | trim) | default(base_zone, true) }}"
+
+    - name: Render GPU MachineSet manifest
+      template:
+        src:  templates/gpu-machineset-azure.j2
+        dest: /tmp/gpu-machineset-azure.yaml
+      vars:
+        ms_name: "nvidia-worker-{{ azure_location | replace(' ', '') }}{{ gpu_zone }}"
+
+    - name: Apply the GPU MachineSet
+      kubernetes.core.k8s:
+        state: present
+        src: /tmp/gpu-machineset-azure.yaml

ansible/playbooks/create-gpu-machine-set.yaml → ansible/playbooks/create-gpu-machineset.yaml

@@ -72,7 +72,7 @@
 
     - name: "[create-gpu-machine-set] Generate machineset"
       ansible.builtin.template:
-        src: templates/gpu-machine-sets.j2
+        src: templates/gpu-machineset.j2
         dest: /tmp/gpu-machineset.yaml
 
     - name: "[create-gpu-machine-set] Apply machineset to cluster {{ clusterId }}"

ansible/playbooks/templates/gpu-machineset-azure.j2

@@ -0,0 +1,44 @@
+apiVersion: machine.openshift.io/v1beta1
+kind: MachineSet
+metadata:
+  name: {{ ms_name }}
+  namespace: openshift-machine-api
+  labels:
+    machine.openshift.io/cluster-api-cluster: {{ cluster_id }}
+spec:
+  replicas: {{ gpu_replicas }}
+  selector:
+    matchLabels:
+      machine.openshift.io/cluster-api-cluster: {{ cluster_id }}
+      machine.openshift.io/cluster-api-machineset: {{ ms_name }}
+  template:
+    metadata:
+      labels:
+        machine.openshift.io/cluster-api-cluster: {{ cluster_id }}
+        machine.openshift.io/cluster-api-machine-role: worker
+        machine.openshift.io/cluster-api-machine-type: worker
+        machine.openshift.io/cluster-api-machineset: {{ ms_name }}
+    spec:
+      taints:
+        - key: odh-notebook
+          value: "true"
+          effect: NoSchedule
+      metadata:
+        labels:
+          node-role.kubernetes.io/odh-notebook: ''
+      providerSpec:
+        value:
+          apiVersion: machine.openshift.io/v1beta1
+          kind: AzureMachineProviderSpec
+          credentialsSecret: {{ cred_secret | to_json }}
+          location: {{ azure_location | to_json }}
+          zone: {{ gpu_zone | to_json }}
+          resourceGroup: {{ resource_group | to_json }}
+          networkResourceGroup: {{ network_resource_group | to_json }}
+          vnet: {{ vnet | to_json }}
+          subnet: {{ subnet | to_json }}
+          vmSize: {{ gpu_vm_size | to_json }}
+          image: {{ image | to_json }}
+          osDisk: {{ os_disk | to_json }}
+          publicIP: {{ public_ip | to_json }}
+          userDataSecret: {{ user_data | to_json }}

ansible/playbooks/templates/gpu-machine-sets.j2 → ansible/playbooks/templates/gpu-machineset.j2

@@ -11,7 +11,7 @@ metadata:
   name: {{ clusterId }}-gpu-{{ cloudRegion }}
   namespace: openshift-machine-api
 spec:
-  replicas: 3
+  replicas: 1
   selector:
     matchLabels:
       machine.openshift.io/cluster-api-cluster: {{ clusterId }}