Add create cserve deployment example script (#91)

* CServe SDK APIs to use cserve_v2
* Add get_cluster_id
* Add example script for create cserve deployment

Author: anandj91

centml/cli/cluster.py

@@ -177,10 +177,13 @@ def get(type, id):
             click.echo(
                 tabulate(
                     [
-                        ("Hugging face model", deployment.model),
+                        ("Hugging face model", deployment.recipe.model),
                         (
                             "Parallelism",
-                            {"tensor": deployment.tensor_parallel_size, "pipeline": deployment.pipeline_parallel_size},
+                            {
+                                "tensor": deployment.recipe.additional_properties['tensor_parallel_size'],
+                                "pipeline": deployment.recipe.additional_properties['pipeline_parallel_size'],
+                            },
                         ),
                         ("Replicas", {"min": deployment.min_scale, "max": deployment.max_scale}),
                         ("Max concurrency", deployment.concurrency or "None"),

centml/sdk/api.py

@@ -6,7 +6,7 @@
     DeploymentStatus,
     CreateInferenceDeploymentRequest,
     CreateComputeDeploymentRequest,
-    CreateCServeDeploymentRequest,
+    CreateCServeV2DeploymentRequest,
 )
 
 from centml.sdk import auth
@@ -32,16 +32,16 @@ def get_compute(self, id):
         return self._api.get_compute_deployment_deployments_compute_deployment_id_get(id)
 
     def get_cserve(self, id):
-        return self._api.get_cserve_deployment_deployments_cserve_deployment_id_get(id)
+        return self._api.get_cserve_v2_deployment_deployments_cserve_v2_deployment_id_get(id)
 
     def create_inference(self, request: CreateInferenceDeploymentRequest):
         return self._api.create_inference_deployment_deployments_inference_post(request)
 
     def create_compute(self, request: CreateComputeDeploymentRequest):
         return self._api.create_compute_deployment_deployments_compute_post(request)
 
-    def create_cserve(self, request: CreateCServeDeploymentRequest):
-        return self._api.create_cserve_deployment_deployments_cserve_post(request)
+    def create_cserve(self, request: CreateCServeV2DeploymentRequest):
+        return self._api.create_cserve_v2_deployment_deployments_cserve_v2_post(request)
 
     def _update_status(self, id, new_status):
         status_req = platform_api_python_client.DeploymentStatusRequest(status=new_status)
@@ -67,8 +67,16 @@ def get_hardware_instances(self, cluster_id=None):
     def get_prebuilt_images(self, depl_type: DeploymentType):
         return self._api.get_prebuilt_images_prebuilt_images_get(type=depl_type)
 
-    def get_cserve_recipe(self):
-        return self._api.get_cserve_recipe_deployments_cserve_recipes_get().results
+    def get_cserve_recipe(self, model=None, hf_token=None):
+        return self._api.get_cserve_recipe_deployments_cserve_recipes_get(model=model, hf_token=hf_token).results
+
+    def get_cluster_id(self, hardware_instance_id):
+        filtered_hw = list(filter(lambda h: h.id == hardware_instance_id, self.get_hardware_instances()))
+
+        if len(filtered_hw) == 0:
+            raise Exception(f"Invalid hardware instance id {hardware_instance_id}")
+
+        return filtered_hw[0].cluster_id
 
 
 @contextmanager

examples/sdk/create_cserve.py

@@ -0,0 +1,34 @@
+import time
+import centml
+from centml.sdk.api import get_centml_client
+from centml.sdk import DeploymentType, CreateCServeV2DeploymentRequest
+
+with get_centml_client() as cclient:
+    # Get fastest recipe for the Qwen model
+    fastest = cclient.get_cserve_recipe(model="Qwen/Qwen2-VL-7B-Instruct")[0].fastest
+
+    # Modify the recipe if necessary
+    fastest.recipe.additional_properties["max_num_seqs"] = 512
+
+    # Create CServeV2 deployment
+    request = CreateCServeV2DeploymentRequest(
+        name="qwen-fastest",
+        cluster_id=cclient.get_cluster_id(fastest.hardware_instance_id),
+        hardware_instance_id=fastest.hardware_instance_id,
+        recipe=fastest.recipe,
+        min_scale=1,
+        max_scale=1,
+        env_vars={},
+    )
+    response = cclient.create_cserve(request)
+    print("Create deployment response: ", response)
+
+    # Get deployment details
+    deployment = cclient.get_cserve(response.id)
+    print("Deployment details: ", deployment)
+
+    # Pause the deployment
+    cclient.pause(deployment.id)
+
+    # Delete the deployment
+    cclient.delete(deployment.id)