使用Couchbase Python SDK时Docker环境出现std::bad_cast错误

本地运行Couchbase相关Python代码完全正常，但构建Docker镜像后启动时，导入Couchbase模块阶段抛出std::bad_cast异常。以下是相关代码、配置及错误信息：

Python SDK代码 #

from config import config
import base64
from couchbase.cluster import Cluster
from couchbase.options import (ClusterOptions,
                               ClusterTimeoutOptions,
                               QueryOptions,
                               WaitUntilReadyOptions)
from couchbase.auth import PasswordAuthenticator
from datetime import timedelta
from couchbase.diagnostics import ServiceType

def query_couchbase(n1ql_query):
        conn_str = config.CLUSTER_CONN_STR
        cb_username = config.CB_USERNAME
        cb_password = config.CB_PASSWORD
        cb_password = base64.b64decode(cb_password.encode('utf-8')).decode('utf-8')
        bucket_name = config.BUCKET_NAME
        auth = PasswordAuthenticator(cb_username, cb_password)
        timeout_opts = ClusterTimeoutOptions(connect_timeout=timedelta(seconds=20),
                                            kv_timeout=timedelta(seconds=20))
        options = ClusterOptions(auth, timeout_options=timeout_opts)
        options = ClusterOptions(auth)
        cluster = Cluster.connect(conn_str, options)
        cluster.wait_until_ready(timedelta(seconds=10),
                                WaitUntilReadyOptions(service_types=[ServiceType.KeyValue, ServiceType.Query]))
        cluster.bucket(bucket_name)
        query_result = cluster.query(n1ql_query)
        return query_result

requirements.txt #

pyJWT~=2.6.0
requests~=2.28.1
urllib3==1.26.6
zipfile36
fastapi~=0.88.0
promise~=2.3
azure-storage-blob==12.14.1
uvicorn
starlette~=0.22.0
httpx~=0.23.1
pydantic~=1.10.2

openai 
num2words 
pandas 
plotly 
tiktoken 
config
python-multipart
nltk
psycopg2-binary

Dockerfile #

FROM python:3.9-slim
WORKDIR /code
RUN apt-get update 
RUN apt-get install -y  \
    libssl1.1 \
    fonts-liberation \
    libasound2 \
    libatk-bridge2.0-0 \
    libatk1.0-0 \
    libatspi2.0-0 \
    libcups2 \
    libdbus-1-3 \
    libdrm2 \
    libgbm1 \
    libgtk-3-0 \
    libnspr4 \
    libnss3 \
    libwayland-client0 \
    libxcomposite1 \
    libxdamage1 \
    libxfixes3 \
    libxkbcommon0 \
    libxrandr2 \
    xdg-utils \
    libu2f-udev \
    libvulkan1
RUN apt-get install --no-install-suggests --no-install-recommends --yes .gyp python3 make g++
COPY ./requirements.txt ./requirements.txt
RUN pip install --upgrade pip
RUN pip install -r ./requirements.txt
RUN pip install -U scikit-learn scipy matplotlib
RUN pip install couchbase
RUN pip install pyOpenSSL
RUN rm -rf .gyp
COPY ./app /code/app
EXPOSE 8080
CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "8080"]

错误信息 #

terminate called after throwing an instance of 'std::bad_cast'
  what():  std::bad_cast

解决方案 #

1. 补充系统依赖并清理无用组件

Couchbase Python SDK底层基于C++实现，slim镜像默认缺少部分运行/编译依赖，且原Dockerfile中安装了大量无用的GUI组件。修改Dockerfile的系统依赖安装部分：

FROM python:3.9-slim
WORKDIR /code

# 安装Couchbase SDK必需的系统依赖，移除无用GUI组件
RUN apt-get update && apt-get install -y --no-install-suggests --no-install-recommends \
    libc6-dev libssl-dev zlib1g-dev \
    python3 make g++ \
    && apt-get clean && rm -rf /var/lib/apt/lists/*

2. 固定Couchbase SDK版本并整合到requirements.txt

在requirements.txt末尾添加指定版本的Couchbase SDK（选择与Python3.9兼容的稳定版）：

couchbase~=4.1.6
pyOpenSSL

同时删除Dockerfile中单独的pip install couchbase和pip install pyOpenSSL命令，统一通过pip install -r requirements.txt安装。

3. 优化Dockerfile构建顺序

调整步骤顺序，减少镜像层并利用缓存：

FROM python:3.9-slim
WORKDIR /code

# 安装系统依赖
RUN apt-get update && apt-get install -y --no-install-suggests --no-install-recommends \
    libc6-dev libssl-dev zlib1g-dev \
    python3 make g++ \
    && apt-get clean && rm -rf /var/lib/apt/lists/*

# 先复制requirements，利用Docker缓存
COPY ./requirements.txt ./requirements.txt
RUN pip install --upgrade pip
RUN pip install -r ./requirements.txt
RUN pip install -U scikit-learn scipy matplotlib

# 复制应用代码
COPY ./app /code/app

EXPOSE 8080
CMD ["uvicorn", "app.main:app", "--host", "0.0.0.0", "--port", "8080"]

问题原因 #

std::bad_cast是C++层面的类型转换异常，在该场景下是因为：

• 原slim镜像缺少Couchbase SDK运行所需的底层系统库，导致SDK初始化时类型转换失败
• 单独安装Couchbase SDK时，编译环境不完整，生成的二进制文件与运行环境不兼容
• 无用的GUI组件可能引入依赖冲突，干扰SDK的正常加载

内容的提问来源于stack exchange，提问作者Adeep Malmotra