Durable Functions部署报错：无法识别语言及Azure存储连接问题

问题描述 #

正在基于Durable Functions构建流程，需并行调用6个函数，全部执行成功后触发存储过程。原流程由Logic Apps实现，因部分函数执行超2分钟触发重启，故迁移至Durable Functions。本地测试（参照微软Python版文档）运行正常，但部署业务函数时遇到以下问题：

1. 无法识别项目语言
2. 指定语言后，仍无法识别Azure Storage连接（已创建临时存储账户，本地与部署环境使用同一连接字符串）

相关代码与配置 #

Function Activity #

def main(url: str) -> str:
    """Trigger the function url"""
    resp = requests.get(url)
    if resp.status_code in (200, 201):
        return True
    return False

Function Orchestrator #

def orchestrator_function(context: DurableOrchestrationContext):
    FUNC_NAMES = os.environ['TBL_FUNCTIONS']

    tasks = []
    for name in FUNC_NAMES:
        tasks.append(context.call_activity('CWFunctionActivity', os.environ[name]))

    results = yield context.task_all(tasks)
    if all(results):
        result = yield context.call_activity('CWFunctionActivity', os.environ['FORECAST_DATASET_PROC'])
    else:
        result = "Check the function execution logs"

    return result

Function HTTP Starter #

async def main(req: HttpRequest, starter: str) -> HttpResponse:
    client = DurableOrchestrationClient(starter)
    instance_id = await client.start_new(req.route_params["functionName"], None, None)

    logging.info(f"Started orchestration with ID = '{instance_id}'.")

    return client.create_check_status_response(req, instance_id)

local.settings.json #

{
    "IsEncrypted": false,
    "Values": {
        "AzureWebJobsStorage": "storage_connection_string",
        "FUNCTIONS_WORKER_RUNTIME": "python",
        "TBL_FUNCTIONS": "['item1', 'item2', 'item3']",
        /*
        "TBL_FUNCTIONS": [
          "TBL_CHARGE_CODES",
          "TBL_EXPENSES",
          "TBL_MEMBERS",
          "TBL_PROJECTS",
          "TBL_PROJECT_TICKETS",
          "TBL_TIME_ENTRIES"
        ],
        */
        "TBL_CHARGE_CODES" : "corresponding_http_url",
        "TBL_EXPENSES": "corresponding_http_url",
        "TBL_MEMBERS" : "corresponding_http_url",
        "TBL_PROJECTS": "corresponding_http_url",
        "TBL_PROJECT_TICKETS": "corresponding_http_url",
        "TBL_TIME_ENTRIES": "corresponding_http_url"
    }
}

问题解决方案 #

1. 修复项目语言识别问题 #

• 确保部署包包含requirements.txt，文件需包含核心依赖：

  azure-functions>=1.17.0
  azure-functions-durable>=1.13.0
  requests>=2.31.0
  

• 在Azure门户的函数应用→配置→应用设置中，确认FUNCTIONS_WORKER_RUNTIME设置为python，并保存配置
• 使用官方工具部署：通过Azure Functions Core Tools执行func azure functionapp publish <app-name>，或VS Code Azure扩展部署，避免手动上传不完整文件

2. 修复Azure Storage连接识别问题 #

• 验证连接字符串：从Azure门户存储账户→访问密钥复制完整连接字符串，确保无空格、截断或格式错误
• 核对应用设置：在Azure门户函数应用→配置→应用设置中，确认AzureWebJobsStorage的值与本地local.settings.json完全一致
• 配置存储权限：为函数应用的系统分配身份（或用户分配身份）授予存储账户的存储队列数据参与者和存储表数据参与者角色
• 检查存储防火墙：若存储账户启用防火墙，需允许函数应用的IP地址或所属虚拟网络访问

代码优化建议 #

Orchestrator环境变量解析修复

当前TBL_FUNCTIONS读取的是字符串格式数组，直接遍历会导致错误，需解析为JSON数组：

import json

def orchestrator_function(context: DurableOrchestrationContext):
    FUNC_NAMES = json.loads(os.environ['TBL_FUNCTIONS'])

    tasks = []
    for name in FUNC_NAMES:
        tasks.append(context.call_activity('CWFunctionActivity', os.environ[name]))

    results = yield context.task_all(tasks)
    if all(results):
        result = yield context.call_activity('CWFunctionActivity', os.environ['FORECAST_DATASET_PROC'])
    else:
        result = "Check the function execution logs"

    return result

同时修改local.settings.json中的TBL_FUNCTIONS为标准JSON数组：

"TBL_FUNCTIONS": ["TBL_CHARGE_CODES", "TBL_EXPENSES", "TBL_MEMBERS", "TBL_PROJECTS", "TBL_PROJECT_TICKETS", "TBL_TIME_ENTRIES"]

内容的提问来源于stack exchange，提问作者Aroosh Rana