refactor(model): Refactor the model download logic, support multiple solutions, and optimize automatic resource deployment

__tests__/e2e/ci-mac-linux.sh

@@ -30,11 +30,18 @@ fi
 echo "test model download"
 cd model
 pip install -r requirements.txt
-export fc_component_function_name=model-$(uname)-$(uname -m)-$RANDSTR
-python deploy_and_test_model.py --model-id iic/cv_LightweightEdge_ocr-recognitoin-general_damo --region cn-shanghai --auto-cleanup
-python deploy_and_test_model.py --model-id Qwen/Qwen2.5-0.5B-Instruct --region cn-shanghai --auto-cleanup
-python deploy_and_test_model.py --model-id iic/cv_LightweightEdge_ocr-recognitoin-general_damo --region cn-shanghai --storage oss --auto-cleanup
-python deploy_and_test_model.py --model-id Qwen/Qwen2.5-0.5B-Instruct --region cn-shanghai --storage oss --auto-cleanup
+export fc_component_function_name=model-$(uname)-$(uname -m)-$RANDSTR-$RANDOM
+python -u deploy_and_test_model.py --model-id iic/cv_LightweightEdge_ocr-recognitoin-general_damo --region cn-shanghai --auto-cleanup
+python -u deploy_and_test_model.py --model-id Qwen/Qwen2.5-0.5B-Instruct --region cn-shanghai --auto-cleanup
+python -u deploy_and_test_model.py --model-id iic/cv_LightweightEdge_ocr-recognitoin-general_damo --region cn-shanghai --storage oss --auto-cleanup
+python -u deploy_and_test_model.py --model-id Qwen/Qwen2.5-0.5B-Instruct --region cn-shanghai --storage oss --auto-cleanup
+
+echo "test model s_file.yaml"
+# python -u test.py
+s model download -t s_file.yaml
+s deploy -y -t s_file.yaml
+s model remove -t s_file.yaml
+s remove -y -t s_file.yaml
 cd ..
 
 echo "test go runtime"

__tests__/e2e/model/deploy_and_test_model.py

@@ -54,7 +54,8 @@ def deploy_model(model_id: str, region: str = "cn-hangzhou", storage: str = "nas
         tuple: (部署的URL, 配置文件路径)
     """
     # 生成函数名称
-    function_name = f"test-{simple_hash(model_id)}"
+    # 加个随机数
+    function_name = f"test-{simple_hash(model_id)}-{secrets.token_hex(4)}"
 
     # 准备请求数据
     deploy_data = {
@@ -177,17 +178,6 @@ def test_model(model_id: str, deploy_url: str, s_yaml_file: str = None):
     model_detail_url = f"{deploy_url}/model/info"
     print(f"正在获取部署后的模型服务详情: {model_detail_url}")
 
-    try:
-        detail_response = requests.get(
-            model_detail_url, headers={"Authorization": f"Bearer {token}"}
-        )
-        if detail_response.status_code == 200:
-            print(f"部署后的模型服务详情: {detail_response.text}")
-        else:
-            print(f"获取部署后的模型服务详情失败: {detail_response.status_code}")
-    except Exception as e:
-        print(f"获取部署后的模型服务详情时出错: {e}")
-
     # 检查是否是vLLM模型（通过配置文件中的启动命令）
     is_vllm_model = False
     if s_yaml_file:
@@ -212,6 +202,22 @@ def test_model(model_id: str, deploy_url: str, s_yaml_file: str = None):
                     print("检测到vLLM模型，将使用专用测试方法")
         except Exception as e:
             print(f"检查模型类型时出错: {e}")
+
+    # 先调用模型详情接口
+    model_detail_url = f"{deploy_url}/model/info"
+    if is_vllm_model:
+        model_detail_url = f"{deploy_url}/v1/models"
+    print(f"正在获取部署后的模型服务详情: {model_detail_url}")
+    try:
+        detail_response = requests.get(
+            model_detail_url, headers={"Authorization": f"Bearer {token}"}
+        )
+        if detail_response.status_code == 200:
+            print(f"部署后的模型服务详情: {detail_response.text}")
+        else:
+            print(f"获取部署后的模型服务详情失败: {detail_response.status_code}")
+    except Exception as e:
+        print(f"获取部署后的模型服务详情时出错: {e}")
 
     if is_vllm_model:
         # 对于vLLM模型，使用专门的测试方法

__tests__/e2e/model/s_file.yaml

@@ -0,0 +1,75 @@
+edition: 3.0.0
+name: ai-model-app
+access: quanxi
+
+vars:
+  region: 'cn-hangzhou'
+
+resources:
+  fc3:
+    component: ${env('fc_component_version', path('../../../'))}
+    type: Function
+    props:
+      logConfig: auto
+      functionName: fc3-model-files-${env('fc_component_function_name', 'nodejs18')}
+      instanceLifecycleConfig:
+        preStop:
+          handler: 'true'
+          timeout: 300
+      gpuConfig:
+        gpuMemorySize: 16384
+        gpuType: fc.gpu.tesla.1
+      nasConfig: auto
+      runtime: custom-container
+      description: test model download files
+      cpu: 8
+      customContainerConfig:
+        image: >-
+          cap-demo-public-registry.cn-hangzhou.cr.aliyuncs.com/cap-app/image-generation-comfyui-agent:v1.1.0-beta.0
+        port: 9000
+      triggers:
+        - triggerConfig:
+            methods:
+              - GET
+              - POST
+              - PUT
+              - PATCH
+              - DELETE
+              - HEAD
+            authType: anonymous
+            disableURLInternet: false
+          triggerName: fc3-model-files-${env('fc_component_function_name', 'nodejs18')}
+          qualifier: LATEST
+          description: http trigger for fc3-model-files-${env('fc_component_function_name', 'nodejs18')}
+          triggerType: http
+      version: 1.6.0
+      timeout: 3600
+      instanceConcurrency: 200
+      diskSize: 61440
+      memorySize: 32768
+      internetAccess: true
+      environmentVariables:
+        BACKEND_TYPE: comfyui
+        MODEL_ASSET_DIR: /mnt/fc3-model-files-${env('fc_component_function_name', 'nodejs18')}
+        REGION: ${vars.region}
+      vpcConfig: auto
+      region: ${vars.region}
+      annotations:
+        modelConfig:
+          solution: funArt
+          source:
+            uri: 'oss://dipper-cache-cn-hangzhou.oss-cn-hangzhou.aliyuncs.com'
+          downloadStrategy:
+            mode: once
+          target:
+            uri: 'nas://auto'
+          files:
+            - source:
+                path: base/comfyui/v0.3.59-alpha
+              target:
+                path: ''
+            - source:
+                path: >-
+                  function-art/comfyui/models/checkpoints/v1-5-pruned-emaonly-fp16.safetensors
+              target:
+                path: models/checkpoints/v1-5-pruned-emaonly-fp16.safetensors

__tests__/e2e/model/test.py

@@ -0,0 +1,109 @@
+#!/usr/bin/env python3
+import os
+import subprocess
+import time
+import sys
+
+def run_command_with_retry(cmd, silent=False, max_retries=3, timeout=300):
+    """执行命令并返回结果，支持重试"""
+    for i in range(max_retries):
+        if not silent:
+            print(f"执行命令: {cmd} (尝试 {i+1}/{max_retries})")
+
+        try:
+            output = subprocess.check_output(cmd, shell=True, stderr=subprocess.STDOUT, text=True, timeout=timeout)
+            return 0, output, ""
+        except subprocess.TimeoutExpired:
+            print(f"命令超时: {cmd}")
+            if i == max_retries - 1:
+                return -1, "", "Command timed out after retries"
+        except Exception as e:
+            print(f"命令执行出错: {str(e)}")
+            if i == max_retries - 1:
+                return -1, "", str(e)
+
+    return -1, "", "Failed after retries"
+
+def main():
+    # 获取函数名（与YAML中一致）
+    fc_component_function_name = os.environ.get('fc_component_function_name', 'nodejs18')
+    function_name = f"fc3-model-files-{fc_component_function_name}"
+
+    print("开始执行模型部署和实例检查流程...")
+    print(f"函数名: {function_name}")
+
+    # 1. 执行模型下载
+    print("1. 执行模型下载: s model download -t s_file.yaml")
+    subprocess.check_output(f"s model download -t s_file.yaml",shell=True)
+
+    # 2. 部署模型
+    print("2. 执行部署: s deploy -y -t s_file.yaml")
+    subprocess.check_output(f"s deploy -y -t s_file.yaml --skip-push",shell=True)
+
+    # 3. 调用函数确保实例启动
+    print("3. 调用函数确保实例启动: s invoke -t s_file.yaml")
+    subprocess.check_output(f"s invoke -t s_file.yaml",shell=True)
+
+    # 4. 等待实例启动
+    print("4. 等待实例启动...")
+    time.sleep(10)
+
+    # 5. 获取实例列表并提取instanceId
+    print("5. 获取实例列表: s instance list -t s_file.yaml")
+    ret_code, instance_output, stderr = run_command_with_retry("s instance list -t s_file.yaml")
+
+    if ret_code == 0:
+        print("实例列表:")
+        print(instance_output)
+
+        # 提取第一个instanceId
+        instance_id = None
+        for line in instance_output.split('\n'):
+            if 'instanceId:' in line:
+                instance_id = line.split('instanceId:')[1].strip()
+                break
+
+        if instance_id:
+            print(f"提取到的instanceId: {instance_id}")
+
+            # 6. 执行详细检查
+            cmd = f"s instance exec --instance-id {instance_id} --cmd 'ls /mnt/{function_name}/models/checkpoints/v1-5-pruned-emaonly-fp16.safetensors'"
+            print(f"6. 执行详细检查: {cmd}")
+            ret_code, find_output, stderr = run_command_with_retry(cmd)
+            print(f"文件查找结果: {find_output}, 错误信息: {stderr}, 状态码: {ret_code}")
+
+            if ret_code == 0:
+                print("文件查找结果:")
+                print(find_output)
+
+                if "v1-5-pruned-emaonly-fp16.safetensors" in find_output:
+                    print("✓ 文件存在路径确认")
+                else:
+                    print("✗ 文件未在/mnt/auto目录下找到")
+                    # 文件不存在，终止流程
+                    sys.exit(1)
+            else:
+                print("文件查找命令执行失败:")
+                print(find_output)
+                print(f"错误信息: {stderr}")
+                # 命令执行失败，终止流程
+                sys.exit(1)
+        else:
+            print("✗ 未找到instanceId")
+    else:
+        print("✗ 获取实例列表失败:")
+        print(instance_output)
+        print(f"错误信息: {stderr}")
+
+    # 7. 执行模型移除
+    print("7. 执行模型移除: s model remove -t s_file.yaml")
+    subprocess.check_output(f"s model remove -t s_file.yaml", shell=True)
+
+    # 8. 移除部署
+    print("8. 移除部署: s remove -y -t s_file.yaml")
+    subprocess.check_output(f"s remove -y -t s_file.yaml", shell=True)
+
+    print("测试流程完成")
+
+if __name__ == "__main__":
+    main()