Docker部署Cassandra+Python操作遇NoHostAvailable/ConnectionRefusedError

问题重现 #

本地通过Docker搭建Cassandra+Python环境，执行docker compose build和docker compose up后，Python应用抛出连接错误：

cassandra.cluster.NoHostAvailable: ('Unable to connect to any servers', {'172.23.0.2:9042': ConnectionRefusedError(111, "Tried connecting to [('172.23.0.2', 9042)]. Last error: Connection refused")})

原项目结构：

cassandra_study/
├── script/
│   └── init-db.cql
├── app.py
├── Dockerfile
└── docker-compose.yml

核心问题分析 #

• Cassandra初始化脚本未自动执行：仅挂载脚本目录，但未触发init-db.cql执行，导致mykeyspace不存在。
• 健康检查配置过于激进：Cassandra启动需数十秒，1秒间隔+10次重试的配置会提前标记服务健康，Python应用过早发起连接。
• Cassandra网络配置缺失：默认配置下容器间通信可能受阻，需显式允许跨地址访问。
• Python连接无重试逻辑：即使健康检查通过，Cassandra可能仍在初始化，直接连接易失败。

修复步骤 #

1. 修改docker-compose.yml #

调整Cassandra的环境变量、健康检查和初始化逻辑，确保服务完全就绪后再启动Python应用：

version: '3'
services:
  python_app:
    build: 
      context: .
      dockerfile: ./Dockerfile
    volumes:
      - ./:/app
    container_name: python_app-container
    depends_on:
      cassandra:
        condition: service_healthy

  cassandra:
    image: cassandra:latest
    volumes:
      - "./script:/scripts"
      - cassandra_data:/var/lib/cassandra  # 持久化数据，避免重启丢失
    ports:
      - 9042:9042
    healthcheck:
      test: ["CMD-SHELL", "cqlsh -e 'DESCRIBE KEYSPACES;' || exit 1"]
      interval: 5s
      timeout: 10s
      retries: 20  # 延长重试次数，给Cassandra足够启动时间
    environment:
      CASSANDRA_LISTEN_ADDRESS: 0.0.0.0  # 允许所有地址访问
      CASSANDRA_RPC_ADDRESS: 0.0.0.0
      CASSANDRA_START_RPC: 'true'
    container_name: cassandra-study-container
    command: >
      bash -c "
        /docker-entrypoint.sh cassandra -f &
        PID=$$!
        until cqlsh -e 'DESCRIBE KEYSPACES;' >/dev/null 2>&1; do
          echo 'Waiting for Cassandra to be ready...'
          sleep 2
        done
        cqlsh -f /scripts/init-db.cql
        wait $$PID
      "

volumes:
  cassandra_data:  # 定义持久化卷

2. 修正init-db.cql #

UUID类型字段不能直接用字符串'1'，需使用合法UUID值或uuid()函数：

CREATE KEYSPACE IF NOT EXISTS mykeyspace WITH replication = {'class': 'SimpleStrategy', 'replication_factor': '1'};

USE mykeyspace;

CREATE TABLE IF NOT EXISTS mytable (
    id UUID PRIMARY KEY,
    message TEXT
);

-- 使用合法UUID值或uuid()函数插入数据
INSERT INTO mytable (id, message) VALUES (uuid(), 'hello i am the first');
INSERT INTO mytable (id, message) VALUES (uuid(), 'hello i am the second');

3. 优化app.py添加连接重试 #

添加重试逻辑，确保Cassandra完全就绪后再执行操作：

from cassandra.cluster import Cluster
import time

def connect_with_retry():
    cluster = None
    session = None
    max_retries = 10
    retry_count = 0
    while retry_count < max_retries:
        try:
            cluster = Cluster(['cassandra'], port=9042)
            session = cluster.connect()
            # 确保keyspace存在
            session.execute("CREATE KEYSPACE IF NOT EXISTS mykeyspace WITH replication = {'class': 'SimpleStrategy', 'replication_factor': '1'}")
            session.set_keyspace('mykeyspace')
            print("Successfully connected to Cassandra")
            return cluster, session
        except Exception as e:
            print(f"Connection attempt {retry_count+1} failed: {e}")
            retry_count += 1
            time.sleep(3)
    raise Exception("Failed to connect to Cassandra after max retries")

cluster, session = connect_with_retry()

try:
    # 插入新数据
    session.execute("INSERT INTO mytable (id, message) VALUES (uuid(), 'ciao mondo sono il terzo')")
    session.execute("INSERT INTO mytable (id, message) VALUES (uuid(), 'ciao mondo sono il quarto')")
    
    # 查询并打印所有数据
    result = session.execute("SELECT * FROM mytable")
    print("\nAll records in mytable:")
    for row in result:
        print(f"ID: {row.id}, Message: {row.message}")
finally:
    cluster.shutdown()

4. 重新部署 #

执行以下命令重新构建并启动服务：

docker compose down -v  # 清理旧容器和卷
docker compose build
docker compose up

验证结果 #

启动后Python应用会输出所有插入的记录，包括init-db.cql中的2条和脚本新增的2条，说明连接和数据操作正常。

内容的提问来源于stack exchange，提问作者user23733163