本地正常但部署后Azure Functions提示ODBC Driver 17找不到

现象 #

• 本地运行Python 3.9.13代码连接Azure数据库完全正常
• 将dev分支部署到Azure Functions后，无论是本地前端连接后端，还是部署的前端连接后端，均触发以下错误：

  ('01000', "[01000] [unixODBC][Driver Manager]Can't open lib 'ODBC Driver 17 for SQL Server' : file not found (0) (SQLDriverConnect)")

• 程序按逻辑重试5次连接后崩溃，返回500错误
• main分支部署到生产环境可正常连接，dev分支未修改数据库连接代码（代码见下方）

排查与解决方案 #

1. 确认Azure Functions运行时的ODBC驱动安装状态 #

Azure Functions的Linux托管环境（尤其是Consumption计划）默认不会预装ODBC Driver 17 for SQL Server——即使requirements.txt包含pyodbc依赖，系统级的驱动仍需手动安装：

• 对比main和dev分支的函数应用宿主环境：如果main使用Windows环境（默认自带ODBC驱动），而dev切换为Linux环境，就会出现这个差异
• 针对Linux环境，添加自定义部署脚本安装驱动：
  在项目根目录创建post_deploy.sh文件，内容如下：

  #!/bin/bash
  apt-get update
  apt-get install -y unixodbc-dev msodbcsql17
  

  然后在Azure Portal的函数应用设置中，开启“启用自定义部署脚本”，确保部署后自动执行该脚本安装驱动

2. 核对部署配置的一致性 #

• 检查dev分支函数应用的AZURE_FUNCTIONS_ENVIRONMENT环境变量是否正确设置，确保非Debug模式下使用ActiveDirectoryMsi身份验证
• 对比main和dev分支的应用设置，确认数据库服务器地址、名称等变量完全一致
• 检查两个分支的部署模板/方式是否相同，比如是否使用了相同的函数应用版本、扩展配置

3. 验证托管标识权限（非Debug模式） #

当DEBUG_MODE=0时，代码使用ActiveDirectoryMsi身份验证，需确认：

• dev分支的函数应用已启用系统托管标识
• 该托管标识已被添加到Azure SQL数据库的IAM角色中（如SQL DB Contributor），或在SQL Server中创建了对应的登录名和用户权限

4. 检查驱动名称格式 #

Linux环境下ODBC驱动名称的大括号可以保留，但需确保驱动实际安装。如果安装后仍报错，可以尝试将驱动名称改为：

driver = "ODBC Driver 17 for SQL Server"  # 去掉大括号

用户提供的数据库连接代码 #

def connect_to_database(server, database, driver, DEBUG_MODE):
        logging.info("ENTERING connect_to_database")
        logging.info(f"Driver: {driver}")
        logging.info(f"Database: {database}")
        logging.info(f"Server: {server}")
        conn = None

        retry_flag = True
        retry_count = 0
        connection_string = (
            f"Driver={driver};Server=tcp:{server};PORT=xxxx;Database={database};"
            "Encrypt=yes;TrustServerCertificate=no;Connection Timeout=5;"
        )

        if DEBUG_MODE:
            connection_string += "UID=my_mail@mail.com;Authentication=ActiveDirectoryInteractive"
        else:
            connection_string += "Authentication=ActiveDirectoryMsi"

        while retry_flag and retry_count < 5:
            try:
                logging.info("Attempting to connect to database...")
                pyodbc.pooling = True
                conn = pyodbc.connect(connection_string)
                retry_flag = False
                logging.info("Database connection successful.")
            except Exception as e:
                logging.error(f"Retry {retry_count + 1} failed: {e}")
                retry_count += 1
                time.sleep(1)

        if conn:
            logging.info("DB connection OK")
        else:
            logging.error("Failed to connect to the database after retries")

        return conn

    def connect_to_MPAPPS():
        logging.info("ENTERING connect_to_MPAPPS")
        conn = connect_to_MPAPPS_Conn()
        if conn:
            logging.info(f"conn result: {conn}")
            cursor = conn.cursor()
            logging.info("SUCCESSFULLY EXITING connect_to_MPAPPS")
            return cursor
        else:
            logging.error("Failed to obtain a database connection")
            return None

    def connect_to_MPAPPS_Conn():
        logging.info("ENTERING connect_to_MPAPPS_Conn")
        server = "server address"
        logging.info(f"Server: {server}")
        database = "database_name"
        logging.info(f"Database: {database}")
        driver = "{ODBC Driver 17 for SQL Server}"
        logging.info(f"Driver: {driver}")

        DEBUG_MODE = 1 if os.getenv("AZURE_FUNCTIONS_ENVIRONMENT") == "Development" else 0
        conn = connect_to_database(server, database, driver, DEBUG_MODE)
        logging.info(f"conn: {conn}")
        logging.info("SUCCESSFULLY EXITING connect_to_MPAPPS_Conn")
        return conn

内容的提问来源于stack exchange，提问作者Dev_Yonko