手把手教你用Docker Compose在Ubuntu 22.04上部署LangSmith监控平台（含PostgreSQL+Redis+ClickHouse配置）

基于Docker Compose的LangSmith全栈监控平台部署指南在当今AI技术快速迭代的背景下构建可观测的LLM应用开发环境已成为技术团队的刚需。LangSmith作为LangChain生态中的核心监控工具能够帮助开发者实时追踪模型调用、分析性能瓶颈并优化工作流程。本文将详细介绍如何在Ubuntu 22.04系统上使用Docker Compose编排PostgreSQL、Redis和ClickHouse三大核心服务搭建完整的LangSmith监控平台。1. 环境准备与基础配置部署LangSmith前需要确保基础环境满足运行要求。推荐使用物理服务器或云主机避免在资源受限的虚拟环境中运行。硬件最低配置要求CPU4核推荐8核内存16GB推荐32GB存储100GB SSD推荐200GB以上网络千兆网卡对于软件环境我们需要准备以下组件# 更新系统并安装基础工具 sudo apt update sudo apt upgrade -y sudo apt install -y curl git python3 python3-pip # 安装Docker引擎 curl -fsSL https://get.docker.com | sudo sh # 安装Docker Compose插件 sudo apt install -y docker-compose-plugin # 验证安装 docker --version docker compose version配置完成后建议将当前用户加入docker组以避免频繁使用sudosudo usermod -aG docker $USER newgrp docker提示执行上述命令后需要重新登录才能使权限变更生效2. 安全密钥与配置文件生成LangSmith运行需要两类关键密钥API密钥盐值(API Key Salt)和JWT密钥(JWT Secret)。这些密钥用于加密存储和身份验证。生成密钥的命令如下# 生成API Key Salt openssl rand -base64 32 api_key_salt.txt # 生成JWT Secret openssl rand -base64 32 jwt_secret.txt生成的密钥应该妥善保管建议将它们存储在安全的位置。接下来创建LangSmith的配置文件config.env# LangSmith核心配置 LANGCHAIN_ENVproduction API_KEY_SALT$(cat api_key_salt.txt) JWT_SECRET$(cat jwt_secret.txt) # 数据库连接配置 POSTGRES_DBlangsmith POSTGRES_USERlangsmith POSTGRES_PASSWORDstrongpassword # Redis配置 REDIS_PASSWORDredispass123 # ClickHouse配置 CLICKHOUSE_PASSWORDclickhousepass3. Docker Compose编排配置创建docker-compose.yml文件定义完整的服务栈。我们将使用PostgreSQL 14作为主数据库Redis 7作为缓存ClickHouse 24作为分析型数据库。version: 3.8 services: # PostgreSQL服务 postgres: image: postgres:14 container_name: langsmith-postgres environment: POSTGRES_USER: ${POSTGRES_USER} POSTGRES_PASSWORD: ${POSTGRES_PASSWORD} POSTGRES_DB: ${POSTGRES_DB} volumes: - postgres_data:/var/lib/postgresql/data healthcheck: test: [CMD-SHELL, pg_isready -U ${POSTGRES_USER}] interval: 5s timeout: 5s retries: 5 networks: - langsmith-net # Redis服务 redis: image: redis:7 container_name: langsmith-redis command: redis-server --requirepass ${REDIS_PASSWORD} volumes: - redis_data:/data healthcheck: test: [CMD, redis-cli, ping] interval: 5s timeout: 5s retries: 5 networks: - langsmith-net # ClickHouse服务 clickhouse: image: clickhouse/clickhouse-server:24.8 container_name: langsmith-clickhouse environment: CLICKHOUSE_PASSWORD: ${CLICKHOUSE_PASSWORD} volumes: - clickhouse_data:/var/lib/clickhouse - ./clickhouse-config.xml:/etc/clickhouse-server/config.d/custom.xml ports: - 8123:8123 # HTTP接口 - 9000:9000 # 原生接口 networks: - langsmith-net # LangSmith后端服务 langsmith-backend: image: langchain/langsmith-backend:latest container_name: langsmith-backend depends_on: postgres: condition: service_healthy redis: condition: service_healthy clickhouse: condition: service_healthy environment: - API_KEY_SALT${API_KEY_SALT} - JWT_SECRET${JWT_SECRET} - POSTGRES_DATABASE_URIpostgresql://${POSTGRES_USER}:${POSTGRES_PASSWORD}postgres:5432/${POSTGRES_DB} - REDIS_DATABASE_URIredis://:${REDIS_PASSWORD}redis:6379/0 - CLICKHOUSE_HOSTclickhouse - CLICKHOUSE_PASSWORD${CLICKHOUSE_PASSWORD} ports: - 1984:1984 networks: - langsmith-net # LangSmith前端服务 langsmith-frontend: image: langchain/langsmith-frontend:latest container_name: langsmith-frontend depends_on: langsmith-backend: condition: service_started environment: - VITE_BACKEND_URLhttp://langsmith-backend:1984 ports: - 1980:1980 networks: - langsmith-net volumes: postgres_data: redis_data: clickhouse_data: networks: langsmith-net: driver: bridge同时创建ClickHouse的优化配置文件clickhouse-config.xmlyandex profiles default max_memory_usage8000000000/max_memory_usage max_threads8/max_threads background_pool_size16/background_pool_size async_insert1/async_insert async_insert_threads4/async_insert_threads /default /profiles /yandex4. 服务启动与验证配置完成后可以启动整个服务栈# 加载环境变量 export $(grep -v ^# config.env | xargs) # 启动服务 docker compose up -d # 查看服务状态 docker compose ps服务启动后可以通过以下方式验证各组件是否正常运行PostgreSQL验证docker exec -it langsmith-postgres psql -U langsmith -d langsmith -c SELECT version();Redis验证docker exec -it langsmith-redis redis-cli -a redispass123 pingClickHouse验证curl -X POST http://localhost:8123 -d SELECT version()LangSmith后端验证curl -X GET http://localhost:1984/health前端访问 打开浏览器访问http://服务器IP:1980应该能看到LangSmith的登录界面。5. 高级配置与优化5.1 数据库性能调优对于生产环境需要对数据库进行针对性优化。以下是PostgreSQL的推荐配置# 进入PostgreSQL容器 docker exec -it langsmith-postgres bash # 编辑PostgreSQL配置 vi /var/lib/postgresql/data/postgresql.conf建议修改以下参数shared_buffers 4GB effective_cache_size 12GB maintenance_work_mem 1GB work_mem 128MB random_page_cost 1.1 max_connections 200对于ClickHouse我们已经在配置文件中设置了异步插入等优化参数还可以通过以下命令进一步优化# 进入ClickHouse容器 docker exec -it langsmith-clickhouse bash # 执行优化命令 clickhouse-client --password clickhousepass --query ALTER TABLE langsmith.runs MODIFY SETTING parts_to_delay_insert1000, parts_to_throw_insert2000; 5.2 容器资源限制为了避免某个服务占用过多资源影响整体稳定性可以在docker-compose.yml中为每个服务添加资源限制services: langsmith-backend: deploy: resources: limits: cpus: 4 memory: 8G reservations: cpus: 2 memory: 4G postgres: deploy: resources: limits: cpus: 4 memory: 6G reservations: cpus: 2 memory: 4G5.3 数据备份策略为确保数据安全需要建立定期备份机制。创建备份脚本backup.sh#!/bin/bash # 备份PostgreSQL docker exec langsmith-postgres pg_dump -U langsmith -Fc langsmith /backups/langsmith_$(date %Y%m%d).dump # 备份Redis docker exec langsmith-redis redis-cli -a redispass123 save docker cp langsmith-redis:/data/dump.rdb /backups/redis_$(date %Y%m%d).rdb # 备份ClickHouse docker exec langsmith-clickhouse clickhouse-backup create docker cp langsmith-clickhouse:/var/lib/clickhouse/backup /backups/clickhouse_$(date %Y%m%d)设置定时任务每天执行备份crontab -e添加以下内容0 2 * * * /path/to/backup.sh6. 常见问题排查6.1 服务启动失败如果某个容器无法启动首先检查日志docker compose logs 服务名常见问题及解决方案端口冲突修改docker-compose.yml中的端口映射内存不足增加服务器内存或调整容器资源限制配置错误检查环境变量和配置文件是否正确6.2 数据不显示如果LangSmith界面没有显示追踪数据检查ClickHouse连接curl -X POST http://localhost:8123 -d SHOW DATABASES后端日志docker compose logs -f langsmith-backend网络连通性docker exec langsmith-backend ping clickhouse docker exec langsmith-backend ping postgres6.3 性能优化建议如果系统响应缓慢可以考虑增加Redis缓存大小优化ClickHouse表结构为频繁查询添加数据库索引调整LangSmith的批处理参数7. 系统维护与监控为了保持系统稳定运行建议部署监控系统。以下是使用cAdvisor和Prometheus的配置示例在docker-compose.yml中添加services: cadvisor: image: gcr.io/cadvisor/cadvisor:v0.47.0 container_name: cadvisor volumes: - /:/rootfs:ro - /var/run:/var/run:ro - /sys:/sys:ro - /var/lib/docker/:/var/lib/docker:ro ports: - 8080:8080 networks: - langsmith-net prometheus: image: prom/prometheus:latest container_name: prometheus volumes: - ./prometheus.yml:/etc/prometheus/prometheus.yml ports: - 9090:9090 networks: - langsmith-net创建prometheus.yml配置文件global: scrape_interval: 15s scrape_configs: - job_name: docker static_configs: - targets: [cadvisor:8080] - job_name: langsmith static_configs: - targets: [langsmith-backend:1984]启动后可以通过http://服务器IP:9090访问Prometheus监控界面。