云函数调用Oracle数据库报DPI-1047错误，如何关联云存储桶中的Oracle Client库？

构建调用Oracle数据库的云函数时遇到以下问题：

• 已编写Python查询代码并存入代码库，HTTP触发器运行正常
• Oracle Client Library已上传至同区域云存储桶，但云函数运行时抛出找不到Oracle客户端库的错误
• 错误信息：

Error: DPI-1047: Cannot locate a 64-bit Oracle Client library: "libclntsh.so: cannot open shared object file: No such file or directory"

相关代码如下：

import cx_Oracle
def queryOracleDatabase(request):
    # Oracle Database Connection
    username = 'x'
    password = 'y'
    connStr = '00.00.00.00:0000/abcd'
    try:
        conn = cx_Oracle.connect(username, password, connStr)
    except cx_Oracle.DatabaseError as e:
        error = e.args
        print('Error: ', error.message)
        return

    # Execute the query
    try:
        cursor = conn.cursor()
        cursor.execute('select * table')  # 注意：此处SQL语法错误，应为select * from table
        data = cursor.fetchall()
    except cx_Oracle.DatabaseError as e:
        error = e.args
        print('Error: ', error.message)
        return
    
    # Clean up
    cursor.close()
    conn.close()
    
    return data

1. 配置云函数的存储桶访问权限 #

为云函数关联的服务账号添加存储对象查看器（或等价的读取权限），确保云函数有权限读取存储桶中的Oracle Client Library文件。

2. 在云函数启动时下载客户端库到临时目录 #

云函数运行环境仅/tmp目录可写，需在代码执行前将存储桶中的Oracle Client Library（建议打包为zip压缩包）下载到该目录并解压。以GCP云函数为例，示例代码片段如下：

from google.cloud import storage
import os
import zipfile

def download_oracle_client():
    bucket_name = "你的存储桶名称"
    source_blob_name = "oracle_client.zip"  # 存储桶中客户端库压缩包的路径
    destination_file_name = "/tmp/oracle_client.zip"
    extract_path = "/tmp/oracle_client"

    # 初始化存储客户端
    storage_client = storage.Client()
    bucket = storage_client.bucket(bucket_name)
    blob = bucket.blob(source_blob_name)

    # 下载压缩包到临时目录
    blob.download_to_filename(destination_file_name)

    # 解压压缩包
    with zipfile.ZipFile(destination_file_name, 'r') as zip_ref:
        zip_ref.extractall(extract_path)
    
    # 设置环境变量，让cx_Oracle能找到客户端库文件
    os.environ['LD_LIBRARY_PATH'] = f"{extract_path}/lib:{os.environ.get('LD_LIBRARY_PATH', '')}"

3. 修改原函数，整合客户端初始化逻辑 #

在queryOracleDatabase函数开头调用上述初始化函数，确保连接Oracle前已完成客户端库的配置：

import cx_Oracle
from google.cloud import storage
import os
import zipfile

def download_oracle_client():
    bucket_name = "你的存储桶名称"
    source_blob_name = "oracle_client.zip"
    destination_file_name = "/tmp/oracle_client.zip"
    extract_path = "/tmp/oracle_client"

    storage_client = storage.Client()
    bucket = storage_client.bucket(bucket_name)
    blob = bucket.blob(source_blob_name)
    blob.download_to_filename(destination_file_name)

    with zipfile.ZipFile(destination_file_name, 'r') as zip_ref:
        zip_ref.extractall(extract_path)
    
    os.environ['LD_LIBRARY_PATH'] = f"{extract_path}/lib:{os.environ.get('LD_LIBRARY_PATH', '')}"

def queryOracleDatabase(request):
    # 先初始化Oracle客户端库
    try:
        download_oracle_client()
    except Exception as e:
        print('Failed to initialize Oracle client: ', str(e))
        return

    # Oracle Database Connection
    username = 'x'
    password = 'y'
    connStr = '00.00.00.00:0000/abcd'
    try:
        conn = cx_Oracle.connect(username, password, connStr)
    except cx_Oracle.DatabaseError as e:
        error = e.args[0] if e.args else e
        print('Connection Error: ', error.message)
        return

    # Execute the query
    try:
        cursor = conn.cursor()
        cursor.execute('select * from table')  # 修正SQL语法错误
        data = cursor.fetchall()
    except cx_Oracle.DatabaseError as e:
        error = e.args[0] if e.args else e
        print('Query Error: ', error.message)
        return
    finally:
        # 确保资源被正确释放
        if 'cursor' in locals():
            cursor.close()
        if 'conn' in locals():
            conn.close()
    
    return data

4. 添加依赖包 #

在云函数的requirements.txt中添加所需依赖：

cx_Oracle>=8.0
google-cloud-storage>=2.0

内容的提问来源于stack exchange，提问作者Gojoe