chore: initialize local deployment baseline
This commit is contained in:
commit
9e2196d69f
30 files changed
+1436
No files matched your search
@@ -0,0 +1,4 @@
|
||||
*.sh text eol=lf
|
||||
*.yml text eol=lf
|
||||
*.yaml text eol=lf
|
||||
.env.example text eol=lf
|
||||
+19
@@ -0,0 +1,19 @@
|
||||
.DS_Store
|
||||
.env
|
||||
.env.*
|
||||
!.env.example
|
||||
|
||||
secrets/
|
||||
certs/
|
||||
*.key
|
||||
*.pem
|
||||
*.p12
|
||||
|
||||
results/raw/*
|
||||
!results/raw/.gitkeep
|
||||
results/tmp/
|
||||
|
||||
*.log
|
||||
*.pid
|
||||
*.swp
|
||||
|
||||
@@ -0,0 +1,66 @@
|
||||
# AI SoulMate 本地部署与硬件适配
|
||||
|
||||
## 从这里开始
|
||||
|
||||
先阅读 [Ubuntu 22.04 本地部署运行手册](docs/DEPLOYMENT-RUNBOOK.md),再阅读 [源码版本核对](docs/SOURCE-RECONCILIATION.md)。当前可以启动 PostgreSQL、Redis 和 MinIO 基础设施 POC;业务系统仍需完成对象存储、管理端 AI-only 模式和生产镜像改造后再编排。
|
||||
|
||||
本仓库是 AI SoulMate 私有化部署的编排与留档仓库,用于固定源码版本、记录硬件基线、建设部署配置并沉淀压测结果。业务源码仍保留在各自仓库中,不复制源码、不改写原有 Git 历史。
|
||||
|
||||
## 当前结论
|
||||
|
||||
- 当前 HP ZCentral 4R 样机适合做单机适配、功能跑通和容量边界测试,不作为生产高可用样板。
|
||||
- 第一阶段在中心服务器部署平台后端、平台前端、AI SoulMate 管理端、PostgreSQL、Redis、本地对象存储和统一网关。
|
||||
- Tripo 暂时继续走外部 API,因此这是“业务数据和基础设施本地化”,不是完全离线部署。
|
||||
- 农场端 `ai-soulmate-client` 暂不部署到中心服务器,后续放到靠近打印机的工作站。
|
||||
- 网关能力是必需的,但现有 `nginx-server` 的公网域名和固定路由不能原样交付,需要在本仓库形成私有部署配置。
|
||||
|
||||
## 仓库边界
|
||||
|
||||
| 内容 | 保存位置 |
|
||||
| --- | --- |
|
||||
| Python/Vue 平台源码 | `ai-soulmate-server` |
|
||||
| 农场端 Spring Boot/Vue 源码 | `ai-soulmate-client` |
|
||||
| 管理端源码 | `mlkj-admin` |
|
||||
| 部署编排、版本锁定、脚本和文档 | 本仓库 |
|
||||
| 小型压测报告和容量结论 | `results/reports/` |
|
||||
| 大型原始日志 | `results/raw/`,默认不进入 Git |
|
||||
| 密钥、证书和真实密码 | 不进入 Git |
|
||||
|
||||
## 目录
|
||||
|
||||
- `docs/`:架构、改造路线、样机基线和压测方案。
|
||||
- `docs/CROSS-COMPUTER-HANDOFF.md`:换电脑继续工作的状态、重建 Git 和接管清单。
|
||||
- `decisions/`:影响交付方式的架构决策。
|
||||
- `inventory/`:参与部署的源码版本清单。
|
||||
- `compose/`:后续形成可交付的 Compose 编排;当前仅保留准入说明。
|
||||
- `scripts/`:硬件与运行状态采集脚本。
|
||||
- `benchmark/`:压测场景和单次运行元数据模板。
|
||||
- `results/`:压测报告与不入库的原始数据。
|
||||
|
||||
## 推荐执行顺序
|
||||
|
||||
1. 固化样机硬件、网络、系统和源码版本基线。
|
||||
2. 完成对象存储抽象、管理端 AI-only 模式和生产镜像改造。
|
||||
3. 建设单机 Compose、初始化/迁移、备份恢复和健康检查。
|
||||
4. 在样机上完成联网 POC,验证 Tripo 出站访问和全业务链路。
|
||||
5. 先测本地链路容量,再单独测包含 Tripo 的端到端体验。
|
||||
6. 增加农场工作站,验证任务下发、断线恢复和打印机适配。
|
||||
7. 根据重复压测数据形成不同客户硬件档位,不凭 CPU/内存规格直接承诺并发量。
|
||||
|
||||
详细任务见 [改造路线](docs/IMPLEMENTATION-ROADMAP.md)。
|
||||
|
||||
## 样机首次采集
|
||||
|
||||
在 Ubuntu 样机上执行:
|
||||
|
||||
```bash
|
||||
./scripts/collect-host-facts.sh > results/raw/host-facts-$(date +%Y%m%d-%H%M%S).txt
|
||||
```
|
||||
|
||||
采集结果不包含 IP、MAC、磁盘序列号和机器 UUID。原始文件默认不入 Git;评审后将脱敏结论写入 `results/reports/`。
|
||||
|
||||
## 版本规则
|
||||
|
||||
- 每次可复现部署必须记录三个源码提交、容器镜像 digest、配置版本和数据库迁移版本。
|
||||
- 建议用 `poc-YYYYMMDD.N`、`rc-X.Y.Z`、`release-X.Y.Z` 标记部署仓库版本。
|
||||
- 本仓库只保存示例变量名;真实值由现场 `.env`、密钥文件或客户密钥系统提供。
|
||||
@@ -0,0 +1,52 @@
|
||||
run_id: "YYYYMMDD-HHMM-scenario-sequence"
|
||||
purpose: ""
|
||||
operator: ""
|
||||
started_at: ""
|
||||
ended_at: ""
|
||||
|
||||
hardware_profile:
|
||||
profile_id: "hp-zcentral-4r-w2225-32g-single-nvme"
|
||||
cpu_limit_changes: "none"
|
||||
memory_gib: 32
|
||||
storage_layout: "single-nvme"
|
||||
network_link_mbps: null
|
||||
|
||||
software:
|
||||
deployment_commit: ""
|
||||
server_commit: "24769465d3ac2cb840c5266c8c0def2aacb641aa"
|
||||
admin_commit: "40bd686128010fbbc6c33f6156a7a3cda5336b3a"
|
||||
client_commit: null
|
||||
image_digests: {}
|
||||
database_migration: ""
|
||||
|
||||
workload:
|
||||
scenario_id: ""
|
||||
dataset_version: ""
|
||||
concurrency: 1
|
||||
duration_seconds: 900
|
||||
request_mix: {}
|
||||
object_sizes: []
|
||||
includes_tripo: false
|
||||
|
||||
configuration:
|
||||
application_profile: ""
|
||||
container_limits: {}
|
||||
database_pool_size: null
|
||||
worker_count: null
|
||||
changed_parameters: []
|
||||
|
||||
result:
|
||||
requests_total: null
|
||||
throughput_per_second: null
|
||||
success_rate_percent: null
|
||||
p50_ms: null
|
||||
p95_ms: null
|
||||
p99_ms: null
|
||||
bottleneck: ""
|
||||
valid: false
|
||||
invalid_reason: ""
|
||||
|
||||
artifacts:
|
||||
raw_directory: ""
|
||||
report_path: ""
|
||||
|
||||
@@ -0,0 +1,24 @@
|
||||
COMPOSE_PROJECT_NAME=ai-soulmate-infra
|
||||
DOCKER_NETWORK_NAME=ai-soulmate-internal
|
||||
TZ=Asia/Shanghai
|
||||
|
||||
POSTGRES_IMAGE=postgres:16-alpine
|
||||
POSTGRES_DB=ai_soulmate
|
||||
POSTGRES_USER=ai_soulmate
|
||||
POSTGRES_PASSWORD=replace-with-a-random-value
|
||||
POSTGRES_HOST_PORT=5432
|
||||
POSTGRES_VOLUME_NAME=ai-soulmate-postgres-data
|
||||
|
||||
REDIS_IMAGE=redis:7-alpine
|
||||
REDIS_PASSWORD=replace-with-a-random-value
|
||||
REDIS_HOST_PORT=6379
|
||||
REDIS_VOLUME_NAME=ai-soulmate-redis-data
|
||||
|
||||
MINIO_IMAGE=minio/minio:latest
|
||||
MINIO_MC_IMAGE=minio/mc:latest
|
||||
MINIO_ROOT_USER=replace-with-a-unique-admin-name
|
||||
MINIO_ROOT_PASSWORD=replace-with-a-random-value
|
||||
MINIO_BUCKET=ai-soulmate
|
||||
MINIO_API_HOST_PORT=9000
|
||||
MINIO_CONSOLE_HOST_PORT=9001
|
||||
MINIO_VOLUME_NAME=ai-soulmate-minio-data
|
||||
@@ -0,0 +1,23 @@
|
||||
# Compose 状态
|
||||
|
||||
当前只提交基础设施 POC:PostgreSQL、Redis 和 MinIO。它不启动业务前端、后端、管理端或网关,因为这些服务还没有完成本地对象存储、管理端 AI-only 模式和生产镜像改造。
|
||||
|
||||
## 运行基础设施
|
||||
|
||||
cd /srv/ai-soulmate-local-deployment
|
||||
./scripts/create-local-env.sh compose/.env
|
||||
docker compose --env-file compose/.env -f compose/docker-compose.infra.yml config
|
||||
docker compose --env-file compose/.env -f compose/docker-compose.infra.yml up -d
|
||||
docker compose --env-file compose/.env -f compose/docker-compose.infra.yml ps
|
||||
./scripts/verify-infra.sh compose/.env
|
||||
|
||||
compose/.env 只放在服务器,已经被 .gitignore 忽略。生成后执行 chmod 600 compose/.env。数据库、Redis 和 MinIO 的宿主机端口只绑定到 127.0.0.1,业务容器接入同一个 Docker 网络后使用服务名访问;不要把这些端口改成 0.0.0.0。
|
||||
|
||||
## 当前门槛
|
||||
|
||||
- ai-soulmate-server 当前 Dockerfile 只有开发目标,开发 Compose 还使用源码 bind mount,不能作为生产镜像。
|
||||
- 平台和管理端仍需把 Alibaba OSS 收敛到可切换的存储接口,并通过本地 S3 兼容服务完成预签名、分片、CORS、权限、备份和恢复测试。
|
||||
- mlkj-admin 仍需实现 AI SoulMate-only 启动模式,避免初始化无关业务数据库和默认账号。
|
||||
- nginx-server 的公网路由不能原样复用;完成业务镜像后再使用环境无关的网关模板或客户现有网关。
|
||||
|
||||
上述门槛完成并记录镜像 digest、迁移版本和恢复演练后,才能新增可交付的应用 Compose。
|
||||
@@ -0,0 +1,123 @@
|
||||
name: ${COMPOSE_PROJECT_NAME:-ai-soulmate-infra}
|
||||
|
||||
services:
|
||||
postgres:
|
||||
image: ${POSTGRES_IMAGE:?POSTGRES_IMAGE is required}
|
||||
restart: unless-stopped
|
||||
environment:
|
||||
POSTGRES_DB: ${POSTGRES_DB:?POSTGRES_DB is required}
|
||||
POSTGRES_USER: ${POSTGRES_USER:?POSTGRES_USER is required}
|
||||
POSTGRES_PASSWORD: ${POSTGRES_PASSWORD:?POSTGRES_PASSWORD is required}
|
||||
TZ: ${TZ:-Asia/Shanghai}
|
||||
ports:
|
||||
- "127.0.0.1:${POSTGRES_HOST_PORT:-5432}:5432"
|
||||
volumes:
|
||||
- postgres_data:/var/lib/postgresql/data
|
||||
healthcheck:
|
||||
test: ["CMD-SHELL", "pg_isready -U \"$${POSTGRES_USER}\" -d \"$${POSTGRES_DB}\""]
|
||||
interval: 10s
|
||||
timeout: 5s
|
||||
retries: 6
|
||||
start_period: 20s
|
||||
mem_limit: 2g
|
||||
cpus: 2.0
|
||||
logging:
|
||||
driver: json-file
|
||||
options:
|
||||
max-size: 10m
|
||||
max-file: "3"
|
||||
networks:
|
||||
- internal
|
||||
|
||||
redis:
|
||||
image: ${REDIS_IMAGE:?REDIS_IMAGE is required}
|
||||
restart: unless-stopped
|
||||
command: ["redis-server", "--appendonly", "yes", "--requirepass", "${REDIS_PASSWORD:?REDIS_PASSWORD is required}"]
|
||||
environment:
|
||||
REDIS_PASSWORD: ${REDIS_PASSWORD:?REDIS_PASSWORD is required}
|
||||
TZ: ${TZ:-Asia/Shanghai}
|
||||
ports:
|
||||
- "127.0.0.1:${REDIS_HOST_PORT:-6379}:6379"
|
||||
volumes:
|
||||
- redis_data:/data
|
||||
healthcheck:
|
||||
test: ["CMD-SHELL", "redis-cli -a \"$${REDIS_PASSWORD}\" ping | grep -q PONG"]
|
||||
interval: 10s
|
||||
timeout: 5s
|
||||
retries: 6
|
||||
start_period: 10s
|
||||
mem_limit: 512m
|
||||
cpus: 1.0
|
||||
logging:
|
||||
driver: json-file
|
||||
options:
|
||||
max-size: 10m
|
||||
max-file: "3"
|
||||
networks:
|
||||
- internal
|
||||
|
||||
minio:
|
||||
image: ${MINIO_IMAGE:?MINIO_IMAGE is required}
|
||||
restart: unless-stopped
|
||||
command: ["server", "/data", "--console-address", ":9001"]
|
||||
environment:
|
||||
MINIO_ROOT_USER: ${MINIO_ROOT_USER:?MINIO_ROOT_USER is required}
|
||||
MINIO_ROOT_PASSWORD: ${MINIO_ROOT_PASSWORD:?MINIO_ROOT_PASSWORD is required}
|
||||
TZ: ${TZ:-Asia/Shanghai}
|
||||
ports:
|
||||
- "127.0.0.1:${MINIO_API_HOST_PORT:-9000}:9000"
|
||||
- "127.0.0.1:${MINIO_CONSOLE_HOST_PORT:-9001}:9001"
|
||||
volumes:
|
||||
- minio_data:/data
|
||||
healthcheck:
|
||||
test: ["CMD-SHELL", "curl -fsS http://127.0.0.1:9000/minio/health/live"]
|
||||
interval: 10s
|
||||
timeout: 5s
|
||||
retries: 6
|
||||
start_period: 20s
|
||||
mem_limit: 1g
|
||||
cpus: 2.0
|
||||
logging:
|
||||
driver: json-file
|
||||
options:
|
||||
max-size: 10m
|
||||
max-file: "3"
|
||||
networks:
|
||||
- internal
|
||||
|
||||
minio-init:
|
||||
image: ${MINIO_MC_IMAGE:?MINIO_MC_IMAGE is required}
|
||||
restart: "no"
|
||||
depends_on:
|
||||
minio:
|
||||
condition: service_healthy
|
||||
environment:
|
||||
MINIO_ROOT_USER: ${MINIO_ROOT_USER:?MINIO_ROOT_USER is required}
|
||||
MINIO_ROOT_PASSWORD: ${MINIO_ROOT_PASSWORD:?MINIO_ROOT_PASSWORD is required}
|
||||
MINIO_BUCKET: ${MINIO_BUCKET:?MINIO_BUCKET is required}
|
||||
entrypoint: ["/bin/sh", "-ec"]
|
||||
command:
|
||||
- |
|
||||
mc alias set local http://minio:9000 "$$MINIO_ROOT_USER" "$$MINIO_ROOT_PASSWORD"
|
||||
mc mb --ignore-existing "local/$$MINIO_BUCKET"
|
||||
mem_limit: 256m
|
||||
cpus: 0.5
|
||||
logging:
|
||||
driver: json-file
|
||||
options:
|
||||
max-size: 10m
|
||||
max-file: "3"
|
||||
networks:
|
||||
- internal
|
||||
|
||||
volumes:
|
||||
postgres_data:
|
||||
name: ${POSTGRES_VOLUME_NAME:-ai-soulmate-postgres-data}
|
||||
redis_data:
|
||||
name: ${REDIS_VOLUME_NAME:-ai-soulmate-redis-data}
|
||||
minio_data:
|
||||
name: ${MINIO_VOLUME_NAME:-ai-soulmate-minio-data}
|
||||
|
||||
networks:
|
||||
internal:
|
||||
name: ${DOCKER_NETWORK_NAME:-ai-soulmate-internal}
|
||||
@@ -0,0 +1,22 @@
|
||||
# ADR-0001:部署仓库与业务源码分离
|
||||
|
||||
- 状态:已接受
|
||||
- 日期:2026-08-05
|
||||
|
||||
## 决策
|
||||
|
||||
建立独立的 `ai-soulmate-local-deployment` Git 仓库,保存部署编排、环境模板、版本清单、硬件基线、迁移工具和容量报告。三个业务源码仓库继续独立演进,本仓库通过提交 SHA 和容器镜像 digest 锁定可复现版本。
|
||||
|
||||
## 原因
|
||||
|
||||
- 保留各系统现有历史、权限和发布节奏;
|
||||
- 客户部署只需要一个明确入口,不需要人工拼接多个仓库配置;
|
||||
- 同一套业务源码可以同时支持云部署和私有部署,避免复制出不可维护的代码分叉;
|
||||
- 部署版本可以独立对应某次硬件适配和压测报告。
|
||||
|
||||
## 约束
|
||||
|
||||
- 通用本地化能力应回到对应源码仓库,通过配置或 provider 开关启用。
|
||||
- 本仓库不使用 Git submodule;交付优先使用固定 digest 的镜像,开发期才按清单检出源码。
|
||||
- 密钥、客户数据和大型原始压测日志不进入 Git。
|
||||
|
||||
@@ -0,0 +1,24 @@
|
||||
# ADR-0002:本地对象存储通过兼容层接入
|
||||
|
||||
- 状态:待验证
|
||||
- 日期:2026-08-05
|
||||
|
||||
## 背景
|
||||
|
||||
平台后端和管理端当前直接依赖阿里 OSS SDK。只替换连接地址不能构成本地部署,必须先解耦 SDK 和业务逻辑。
|
||||
|
||||
## 暂定决策
|
||||
|
||||
业务代码使用统一对象存储接口,至少实现 Alibaba OSS 与本地 S3 兼容存储两个 provider。具体本地产品不在完成兼容性测试前锁定;优先评估客户已有 S3 服务与许可证清晰、可离线交付的候选方案。
|
||||
|
||||
## 准入测试
|
||||
|
||||
- 服务端上传/下载/删除、对象存在性和元数据;
|
||||
- 浏览器预签名 PUT/GET、CORS 和过期行为;
|
||||
- 分片上传、大文件、中文文件名和内容类型;
|
||||
- 权限隔离、凭证轮换、审计和容量配额;
|
||||
- 备份、恢复、升级、磁盘满和异常断电恢复;
|
||||
- 与现有 OSS URL/对象数据的迁移和回滚。
|
||||
|
||||
通过准入测试并确认许可证、运维与备份责任后,再将选定实现加入正式 Compose。
|
||||
|
||||
@@ -0,0 +1,58 @@
|
||||
# 本地部署目标架构
|
||||
|
||||
## 第一阶段:中心适配服务器
|
||||
|
||||
```text
|
||||
浏览器 / 管理员
|
||||
|
|
||||
HTTPS 443
|
||||
|
|
||||
统一网关(Nginx 或客户现有网关)
|
||||
|--------------------|
|
||||
平台前端 AI SoulMate 管理端
|
||||
|--------------------|
|
||||
平台后端
|
||||
|----------|----------|
|
||||
PostgreSQL Redis 本地 S3 兼容对象存储
|
||||
|
||||
平台后端 ---- HTTPS 出站 ---- Tripo API
|
||||
```
|
||||
|
||||
除 Tripo 调用外,数据库、缓存、对象文件、平台服务和管理端均位于客户局域网。默认只对业务网开放网关的 `443`;PostgreSQL、Redis 和对象存储管理端口不直接暴露。
|
||||
|
||||
## 第二阶段:农场工作站
|
||||
|
||||
```text
|
||||
中心平台后端 -- HTTPS / 设备身份认证 --> 农场端工作站
|
||||
|
|
||||
本地任务缓存 / 客户端数据库
|
||||
|
|
||||
打印机 / Moonraker
|
||||
```
|
||||
|
||||
农场端主动发起心跳、拉取任务和上报状态,中心端不直接访问打印机网络。后续需要补齐设备级凭证、幂等任务、断网重试、升级和审计能力。
|
||||
|
||||
## 服务职责
|
||||
|
||||
| 服务 | 第一阶段位置 | 说明 |
|
||||
| --- | --- | --- |
|
||||
| `ai-soulmate-server` | 中心服务器 | 前端与 FastAPI 后端 |
|
||||
| `mlkj-admin` AI SoulMate 部分 | 中心服务器 | 必须支持 AI-only 模式,不能强依赖其他业务库 |
|
||||
| PostgreSQL | 中心服务器 | 使用独立数据库/角色,定期备份 |
|
||||
| Redis | 中心服务器 | 只在容器网络开放,启用持久化与认证 |
|
||||
| 本地对象存储 | 中心服务器 | 通过统一存储接口接入,不继续绑定阿里 `oss2` |
|
||||
| 统一网关 | 中心服务器 | TLS、路由、上传限制、访问日志和基础限流 |
|
||||
| Tripo | 外部服务 | 当前为必要出站依赖,指标必须与本地容量分开 |
|
||||
| `ai-soulmate-client` | 后续农场工作站 | 靠近打印机,暂不纳入中心服务器 POC |
|
||||
|
||||
## 网关判断
|
||||
|
||||
网关能力需要保留,但不建议直接部署现有 `nginx-server`:它面向现有公网域名和固定容器名,目前也没有完整覆盖管理端与农场端。私有版应在本仓库维护环境无关的路由模板;如果客户已有负载均衡或企业网关,则由其替代 Nginx。
|
||||
|
||||
## 安全边界
|
||||
|
||||
- 禁止把数据库、Redis 和对象存储管理端口映射到公网。
|
||||
- Tripo、短信、支付等外部能力按现场需求逐项启用,默认关闭非必要集成。
|
||||
- 平台用户、管理员和农场设备使用不同身份体系与凭证。
|
||||
- 所有密码和 API Key 在部署时注入,不构建进镜像、不写入 Git。
|
||||
- 管理端、备份和监控访问应限制到管理网或 VPN。
|
||||
@@ -0,0 +1,66 @@
|
||||
# 容量与压测方案
|
||||
|
||||
## 目标
|
||||
|
||||
压测用于回答“在某个明确硬件、软件版本、配置和业务模型下,能够稳定承载多少负载”,而不是只给出一个脱离场景的并发数。
|
||||
|
||||
必须分开衡量:
|
||||
|
||||
1. 本地平台容量:网关、后端、数据库、Redis 和对象存储。
|
||||
2. 外部依赖体验:Tripo 的排队、生成、轮询、下载、限流和错误。
|
||||
3. 农场端能力:后续工作站接入后的任务吞吐、断线恢复和打印机处理速度。
|
||||
|
||||
## 测试场景
|
||||
|
||||
| 编号 | 场景 | 主要瓶颈 |
|
||||
| --- | --- | --- |
|
||||
| S01 | 健康检查、登录、用户与模型列表读取 | 网关、API、数据库连接池 |
|
||||
| S02 | 管理端搜索、分页、统计查询 | SQL、索引、CPU、数据库缓存 |
|
||||
| S03 | 文件预签名、上传和下载 | 网络、对象存储、磁盘 I/O |
|
||||
| S04 | 任务创建、状态轮询、结果持久化 | API、Redis、数据库写入 |
|
||||
| S05 | Tripo 提交与轮询 | 外部 API 延迟/限流、本地状态机 |
|
||||
| S06 | 读写、文件和任务的业务比例混合负载 | 整机资源争用 |
|
||||
| S07 | 后续中心到农场端任务分发与状态回传 | 设备队列、网络、幂等与重试 |
|
||||
|
||||
S05 必须同时报告“本地处理耗时”和“Tripo 外部等待耗时”,不能用 Tripo 的波动推导本地服务器极限。
|
||||
|
||||
## 负载方法
|
||||
|
||||
- 先做 5 分钟预热,再保持 15–30 分钟稳态,最后观察 5 分钟恢复。
|
||||
- 从低负载开始,以 `1、2、5、10、15、20…` 并发逐级增加;接近瓶颈后缩小步长。
|
||||
- 每个稳态档位重复至少 3 次,报告中位数以及最好/最差值,不只保留一次结果。
|
||||
- 使用固定的脱敏数据集、对象大小分布和业务操作比例。
|
||||
- 文件场景至少区分小文件、典型模型文件和大文件;记录客户端与服务器网络速率。
|
||||
- 在 32GB、64GB 或不同磁盘方案之间对比时,只改变一个主要变量。
|
||||
|
||||
## 必采指标
|
||||
|
||||
- 业务:吞吐量、成功率、错误码、p50/p95/p99、超时、队列长度。
|
||||
- 主机:CPU/Load、可用内存、Swap in/out、磁盘吞吐/IOPS/await/利用率、网络吞吐与丢包。
|
||||
- 容器:CPU、内存、重启次数、健康状态、限额和 OOM。
|
||||
- PostgreSQL:活跃/等待连接、连接池、慢 SQL、锁等待、缓存命中、事务率。
|
||||
- Redis:命令延迟、连接、内存、eviction、持久化状态。
|
||||
- 对象存储:上传/下载吞吐、首字节时间、错误率、磁盘占用。
|
||||
- Tripo:提交/轮询/下载分别耗时,API 限流、错误、外部排队时长。
|
||||
|
||||
## 停止与失效条件
|
||||
|
||||
以下任一情况出现时停止继续加压,保留现场数据并定位瓶颈:
|
||||
|
||||
- 发生 OOM、持续 Swap 抖动、容器反复重启或数据错误;
|
||||
- 错误率持续超过该场景 SLA(初始探索可用 1% 作为警戒线);
|
||||
- p95/p99 持续恶化且负载降低后不能恢复;
|
||||
- CPU 长时间接近满载,或磁盘等待/队列持续增长;
|
||||
- 数据库连接池耗尽、锁等待累积或任务队列只增不减;
|
||||
- 外部 API 进入限流,此时记录为外部边界,不继续推高调用。
|
||||
|
||||
## 单次运行产物
|
||||
|
||||
每次运行必须保存:
|
||||
|
||||
- `benchmark/run-metadata.template.yml` 对应的完整元数据;
|
||||
- 压测脚本版本、开始/结束时间、场景和并发曲线;
|
||||
- 原始请求结果与监控快照(放 `results/raw/`);
|
||||
- 瓶颈分析和客户可读结论(放 `results/reports/`);
|
||||
- 本次结论的适用范围、已知限制和下一步验证。
|
||||
|
||||
@@ -0,0 +1,183 @@
|
||||
# AI SoulMate 本地部署跨电脑交接说明
|
||||
|
||||
## 2026-08-05 本次接管更新
|
||||
|
||||
- 已从用户提供的压缩包恢复部署仓库,确认 Git 分支为 main,但此前没有任何 commit,也没有远端。
|
||||
- 已新增 Ubuntu 22.04 运行手册、源码版本核对说明、Docker 官方仓库安装脚本、随机密钥生成脚本和源码基线采集脚本。
|
||||
- 已新增 PostgreSQL、Redis、MinIO 基础设施 POC Compose;三个宿主机端口只绑定到 127.0.0.1,并带持久化、健康检查、资源上限和日志轮换。
|
||||
- 接管前当前工作区只有另一个指向 Gitee/Spring Boot 的 ai-soulmate-server 本地仓库,不能作为内部 FastAPI 仓库的发布基线。
|
||||
- 已只读核对并浅克隆四个内部远端:server develop6@9a6f7a5、client master@9ebb082、admin main@40bd686、nginx master@6f552f4;server/client 与旧交接基线不同。
|
||||
- server 默认分支使用 Hunyuan3D/Bailian 而不是需求描述的 Tripo,因此发布分支选择仍是业务改造前的阻断项。
|
||||
- 四个仓库存在被 Git 跟踪的真实环境文件,server 还含支付/私钥类值;所有相关凭据必须先轮换,具体值未复制到部署仓库。
|
||||
- 当前 ai-soulmate-server 的 Dockerfile 只有开发目标,开发 Compose 使用源码 bind mount,而且 backend-java/.env 已被源码仓库跟踪;部署前必须审查并轮换其中凭据。
|
||||
- 下一安全动作是在服务器运行 collect-host-facts.sh,完成系统更新和 Docker 安装,然后重新克隆四个源码仓库并生成 source-check 报告。
|
||||
- 首次 Git commit 会在本次接管检查完成后创建;远端仍需用户在 Gitea 建立空的私有仓库后再配置和推送。
|
||||
|
||||
更新时间:2026-08-05
|
||||
|
||||
这份文档用于在另一台电脑继续本地部署适配。它记录事实、已完成事项和下一步顺序;不要把文档中的示例 IP、密码或 API Key 当成生产配置。
|
||||
|
||||
## 一句话状态
|
||||
|
||||
服务器已经可以通过 SSH 访问,但业务系统尚未部署;本地部署 Git 仓库只在当前电脑创建过,尚未建立远端、尚未完成首个提交。换电脑时需要复制或重新建立该仓库。
|
||||
|
||||
## 已确认事实
|
||||
|
||||
### 服务器
|
||||
|
||||
- Ubuntu 22.04.2 LTS,x86_64,内核 5.19.0-32-generic。
|
||||
- 网卡 `enp1s0` 已拿到 `192.168.3.102`,SSH 已成功连接。
|
||||
- `openssh-server` 已安装并可以连接。
|
||||
- `ufw status verbose` 显示 `inactive`,目前没有 UFW 规则阻断 SSH。
|
||||
- 尚未确认 SSH 公钥登录是否完成;尚未关闭密码登录。
|
||||
- 尚未完成系统更新、Docker、Compose、PostgreSQL、Redis、对象存储和网关部署。
|
||||
- 应在路由器或网络设备上为服务器做 DHCP 地址保留,避免重启后 IP 变化。
|
||||
|
||||
### 样机硬件
|
||||
|
||||
- HP ZCentral 4R Workstation。
|
||||
- Intel Xeon W-2225,4 核 8 线程。
|
||||
- 32GiB ECC DDR4,单条内存,约 1TB 单块 Samsung NVMe。
|
||||
- Swap 2GiB,单盘无冗余。
|
||||
- 适合 POC、适配和容量边界测试,不可直接作为生产高可用承诺。
|
||||
|
||||
### 源码版本基线
|
||||
|
||||
| 仓库 | 当前提交 |
|
||||
| --- | --- |
|
||||
| `ai-soulmate-server` | `24769465d3ac2cb840c5266c8c0def2aacb641aa` |
|
||||
| `ai-soulmate-client` | `170847159a787c7b52e3f9f792d72ea67cbd7fa4` |
|
||||
| `mlkj-admin` | `40bd686128010fbbc6c33f6156a7a3cda5336b3a` |
|
||||
| `nginx-server` | `6f552f4b37505ec055cdd179fe9cc5a6825a4189`,仅做过远端检查 |
|
||||
|
||||
完整清单见 [`inventory/source-baselines.yml`](../inventory/source-baselines.yml)。
|
||||
|
||||
## 当前部署目标
|
||||
|
||||
中心服务器先运行:
|
||||
|
||||
```text
|
||||
统一网关 → 平台前端 / AI SoulMate 管理端 → 平台后端
|
||||
├→ PostgreSQL
|
||||
├→ Redis
|
||||
├→ 本地 S3 兼容对象存储
|
||||
└→ Tripo 外部 API
|
||||
```
|
||||
|
||||
后续农场工作站运行 `ai-soulmate-client`,通过 HTTPS 心跳、拉取任务和上报状态,不把打印机直接暴露给中心服务器。
|
||||
|
||||
Tripo 当前仍是外部 API,因此目标是本地化业务平台和数据基础设施,不是当前阶段的完全离线部署。
|
||||
|
||||
## 换电脑时重建 Git 仓库
|
||||
|
||||
### 推荐:复制现有部署目录
|
||||
|
||||
把当前电脑的整个目录复制到新电脑,包含未提交文件但不需要复制 `.git`:
|
||||
|
||||
```text
|
||||
/Users/max/AI SoulMate/开发/ai-soulmate-local-deployment
|
||||
```
|
||||
|
||||
在新电脑进入目录后检查:
|
||||
|
||||
```bash
|
||||
git status
|
||||
git branch --show-current
|
||||
```
|
||||
|
||||
如果没有 `.git`,重新初始化:
|
||||
|
||||
```bash
|
||||
git init -b main
|
||||
```
|
||||
|
||||
### 需要长期留档时
|
||||
|
||||
先在 Gitea 创建一个空的私有仓库,例如 `MLKJ/ai-soulmate-local-deployment`,再在新电脑执行:
|
||||
|
||||
```bash
|
||||
git add .
|
||||
git commit -m "chore: initialize local deployment baseline"
|
||||
git remote add origin <GITEA_REPOSITORY_URL>
|
||||
git push -u origin main
|
||||
```
|
||||
|
||||
远端 URL、账号和访问令牌不要写入文档或 `.env`。如果部署目录没有复制过来,必须先恢复本目录中的 `README.md`、`docs/`、`inventory/`、`benchmark/`、`scripts/`、`compose/` 和 `decisions/`,再执行上述提交。
|
||||
|
||||
## 新电脑接管顺序
|
||||
|
||||
### A. 先固定 SSH 访问
|
||||
|
||||
在新电脑生成或准备 SSH 密钥,把公钥加入服务器 `hp` 用户:
|
||||
|
||||
```bash
|
||||
ssh-keygen -t ed25519 -a 64 -f ~/.ssh/id_ed25519_ai_soulmate
|
||||
cat ~/.ssh/id_ed25519_ai_soulmate.pub | ssh hp@192.168.3.102 \
|
||||
'umask 077; mkdir -p ~/.ssh; cat >> ~/.ssh/authorized_keys'
|
||||
```
|
||||
|
||||
确认密钥登录成功后,才考虑关闭密码登录。修改 SSH 配置前保留当前会话,并先用 `sudo sshd -t` 检查配置。
|
||||
|
||||
### B. 记录并更新服务器
|
||||
|
||||
在服务器上执行:
|
||||
|
||||
```bash
|
||||
whoami
|
||||
hostnamectl
|
||||
ip -br addr
|
||||
sudo systemctl is-active ssh
|
||||
df -hT /
|
||||
free -h
|
||||
lscpu
|
||||
|
||||
sudo timedatectl set-timezone Asia/Shanghai
|
||||
sudo apt update
|
||||
sudo apt install -y git curl ca-certificates gnupg jq unzip rsync tmux htop nvme-cli smartmontools
|
||||
sudo apt upgrade -y
|
||||
```
|
||||
|
||||
如果出现 `/var/run/reboot-required`,记录后再重启,并重新确认服务器 IP。
|
||||
|
||||
### C. 安装容器运行时
|
||||
|
||||
使用 Docker 官方 Ubuntu 仓库安装 Docker Engine、Buildx 和 Compose plugin。安装前确认网络、DNS、NTP 和磁盘空间;不要使用来源不明的一键安装脚本。
|
||||
|
||||
安装后至少验证:
|
||||
|
||||
```bash
|
||||
docker version
|
||||
docker compose version
|
||||
sudo systemctl is-active docker
|
||||
```
|
||||
|
||||
### D. 再开始业务改造
|
||||
|
||||
不要直接把当前四个仓库用现有 Compose 上线。必须先完成:
|
||||
|
||||
1. 平台后端和管理端的 OSS provider 抽象,并验证本地 S3 兼容对象存储。
|
||||
2. 管理端 `ai-soulmate-only` 模式,避免强制连接其他业务数据库。
|
||||
3. 生产前端构建和不可变镜像,去除源码 bind mount。
|
||||
4. 初始化/迁移、备份/恢复、健康检查和密钥注入。
|
||||
5. 网关路由、TLS、上传限制和管理入口隔离。
|
||||
|
||||
具体路线见 [`IMPLEMENTATION-ROADMAP.md`](IMPLEMENTATION-ROADMAP.md),压测规则见 [`BENCHMARK-PLAN.md`](BENCHMARK-PLAN.md)。
|
||||
|
||||
## 不要做的事情
|
||||
|
||||
- 不要把 SSH、PostgreSQL、Redis 或对象存储管理端口暴露到公网。
|
||||
- 不要把真实密码、Tripo Key、OSS Key、证书或客户数据提交到 Git。
|
||||
- 不要把 Tripo 的端到端延迟直接当作本地服务器并发能力。
|
||||
- 不要在单盘、32GB POC 样机上直接承诺生产高可用或客户 SLA。
|
||||
- 不要在没有备份恢复演练前导入唯一生产数据。
|
||||
|
||||
## 接管清单
|
||||
|
||||
- [ ] 新电脑已恢复部署仓库目录。
|
||||
- [ ] Git 仓库已初始化并完成首次提交。
|
||||
- [ ] 已创建私有远端并推送 `main`。
|
||||
- [ ] Mac/新电脑使用密钥登录服务器。
|
||||
- [ ] 服务器 IP 已固定或完成 DHCP 保留。
|
||||
- [ ] 已完成系统更新和硬件基线采集。
|
||||
- [ ] 已安装并验证 Docker Compose。
|
||||
- [ ] 已完成 P1/P2 源码改造后再创建正式 Compose。
|
||||
@@ -0,0 +1,110 @@
|
||||
# Ubuntu 22.04 本地部署运行手册
|
||||
|
||||
这台 HP ZCentral 4R 是 POC 和适配样机。当前目标是跑通中心服务、管理端、PostgreSQL、Redis、本地对象存储和 Tripo 出站调用;它不是生产高可用承诺。
|
||||
|
||||
## 0. 先确认网络和 SSH
|
||||
|
||||
在服务器本地控制台或当前 SSH 会话记录:
|
||||
|
||||
~~~bash
|
||||
hostnamectl
|
||||
nmcli device status
|
||||
ip -br addr
|
||||
sudo systemctl is-active ssh
|
||||
df -hT /
|
||||
free -h
|
||||
~~~
|
||||
|
||||
在路由器上为有线网卡做 DHCP reservation,或者使用网络管理员批准的静态地址。不要凭 enp1s0、eno1 这样的示例名修改连接。保留当前密码 SSH 会话,直到另一台电脑用 Ed25519 密钥成功登录。
|
||||
|
||||
## 1. 固化主机基线
|
||||
|
||||
~~~bash
|
||||
cd /srv/ai-soulmate-local-deployment
|
||||
chmod +x scripts/*.sh
|
||||
./scripts/collect-host-facts.sh \
|
||||
> results/raw/host-facts-$(date +%Y%m%d-%H%M%S).txt
|
||||
~~~
|
||||
|
||||
原始文件默认不进 Git。审阅后只把脱敏结论写入 results/reports/,不要记录 IP、MAC、序列号、UUID 或客户数据。
|
||||
|
||||
## 2. 安装 Docker 和工具
|
||||
|
||||
~~~bash
|
||||
cd /srv/ai-soulmate-local-deployment
|
||||
sudo TIMEZONE_NAME=Asia/Shanghai RUN_UPGRADE=1 \
|
||||
./scripts/bootstrap-ubuntu-22.04.sh
|
||||
~~~
|
||||
|
||||
脚本使用 Docker 官方 apt 仓库,不修改网络地址、不自动打开防火墙端口,也不会替代现场重启决策。若出现 /var/run/reboot-required,先记下结果,再重启并重新确认 IP、SSH、磁盘和 Docker。
|
||||
|
||||
## 3. 放置并核对源码
|
||||
|
||||
源码与部署仓库分开,例如:
|
||||
|
||||
~~~bash
|
||||
sudo mkdir -p /srv/ai-soulmate/sources
|
||||
sudo chown -R "$USER":"$USER" /srv/ai-soulmate
|
||||
cd /srv/ai-soulmate/sources
|
||||
git clone http://39.106.119.229:3000/MLKJ/ai-soulmate-server
|
||||
git clone http://39.106.119.229:3000/MLKJ/ai-soulmate-client
|
||||
git clone http://39.106.119.229:3000/MLKJ/mlkj-admin
|
||||
git clone http://39.106.119.229:3000/MLKJ/nginx-server
|
||||
~~~
|
||||
|
||||
如果仓库已经存在,不要直接覆盖用户改动;逐个执行 git status --short、git remote -v 和 git rev-parse HEAD。完成后生成核对报告:
|
||||
|
||||
截至 2026-08-05,server 默认分支是 develop6,但其 AI provider 与需求描述不一致;client 的旧交接 SHA 也不是默认分支 HEAD。克隆完成不代表发布分支已经选定,必须由仓库负责人确认 release branch/tag 后再继续。
|
||||
|
||||
~~~bash
|
||||
cd /srv/ai-soulmate-local-deployment
|
||||
SOURCE_ROOT=/srv/ai-soulmate/sources \
|
||||
./scripts/record-source-baseline.sh \
|
||||
> inventory/source-check-$(date +%F-%H%M%S).yml
|
||||
~~~
|
||||
|
||||
## 4. 先启动基础设施
|
||||
|
||||
~~~bash
|
||||
cd /srv/ai-soulmate-local-deployment
|
||||
./scripts/create-local-env.sh compose/.env
|
||||
docker compose --env-file compose/.env \
|
||||
-f compose/docker-compose.infra.yml config
|
||||
docker compose --env-file compose/.env \
|
||||
-f compose/docker-compose.infra.yml pull
|
||||
docker compose --env-file compose/.env \
|
||||
-f compose/docker-compose.infra.yml up -d
|
||||
docker compose --env-file compose/.env \
|
||||
-f compose/docker-compose.infra.yml ps
|
||||
./scripts/verify-infra.sh compose/.env
|
||||
~~~
|
||||
|
||||
PostgreSQL、Redis 和 MinIO 只绑定到 127.0.0.1。未来业务容器使用 postgres:5432、redis:6379 和 minio:9000,不要把 .env 中的宿主机端口当作容器间地址。MinIO 控制台需要时使用 SSH 隧道,不开放公网管理端口。
|
||||
|
||||
## 5. 业务部署门槛
|
||||
|
||||
当前 ai-soulmate-server 的 Dockerfile 只有开发目标,开发 Compose 还挂载源码;压缩包中的 ai-soulmate-server-master 没有可验证的 Git 历史。先完成以下改造,再写应用 Compose:
|
||||
|
||||
1. 平台和管理端引入统一对象存储接口,同时保留 Alibaba OSS provider,并让 MinIO 等本地 S3 provider 通过预签名上传、下载、分片、CORS、权限、元数据和恢复测试。
|
||||
2. mlkj-admin 增加 AI SoulMate-only profile,不能初始化无关业务库、默认账号或固定密码。
|
||||
3. 为后端、平台前端和管理端生成生产镜像;禁止源码 bind mount,固定基础镜像版本并记录 digest。
|
||||
4. 添加迁移、健康检查、资源上限、日志轮换、备份恢复和回滚流程。
|
||||
5. 再决定是否使用 nginx-server。它只承担环境无关的 TLS、路由、上传限制、访问日志和限流;若现场已有企业网关,优先复用企业网关。
|
||||
|
||||
## 6. 验收顺序
|
||||
|
||||
按 S01 到 S07 的场景逐级验收:健康检查和登录 → 管理端查询 → 对象上传下载 → 任务创建与轮询 → Tripo 出站 → 混合负载 → 工作站任务下发。Tripo 的提交、轮询、下载、外部排队和限流时间必须单独记录,不能作为本地并发能力。
|
||||
|
||||
每次压测使用固定数据集和请求比例,从并发 1, 2, 5, 10, 15, 20... 逐级增加;每个稳定档位至少重复三次,保存元数据、脱敏报告和原始日志位置。出现 OOM、持续 swap、容器重启、数据错误、队列增长或外部 API 限流时停止加压。
|
||||
|
||||
## 7. Git 留档
|
||||
|
||||
~~~bash
|
||||
cd /srv/ai-soulmate-local-deployment
|
||||
git add .
|
||||
git commit -m "chore: initialize local deployment baseline"
|
||||
git remote add origin <GITEA_REPOSITORY_URL>
|
||||
git push -u origin main
|
||||
~~~
|
||||
|
||||
远端仓库需要先在 Gitea 创建为空的私有仓库。不要把账号密码、访问令牌或真实 .env 写进 URL、文档或 Git;当前部署仓库只有在产生首个 commit 后才算完成留档。
|
||||
@@ -0,0 +1,67 @@
|
||||
# 本地部署改造路线
|
||||
|
||||
## P0:冻结基线
|
||||
|
||||
- [x] 记录现有三个源码仓库提交。
|
||||
- [x] 记录样机 CPU、内存、磁盘、操作系统和网络状态。
|
||||
- [ ] 为四个源码仓库建立本地部署开发分支和对应需求清单。
|
||||
- [ ] 确认数据库版本、现有数据量、OSS 对象量和允许停机窗口。
|
||||
|
||||
验收:任何一次测试都能关联到硬件、源码、配置和数据集版本。
|
||||
|
||||
## P1:平台后端与对象存储解耦
|
||||
|
||||
- [ ] 将阿里 OSS 调用收敛为对象存储接口。
|
||||
- [ ] 实现 Alibaba OSS 与本地 S3 兼容存储两个 provider。
|
||||
- [ ] 验证预签名上传/下载、分片上传、CORS、文件元数据和清理任务。
|
||||
- [ ] 将 Tripo 封装为 provider,保留外部 API,并为未来私有端点预留配置。
|
||||
- [ ] 移除生产 Compose 中的源码 bind mount,构建不可变镜像。
|
||||
|
||||
验收:切换存储 provider 不需要修改业务代码;对象迁移后原有 URL 使用场景可回归。
|
||||
|
||||
## P2:管理端 AI-only 模式
|
||||
|
||||
- [ ] 增加 `ai-soulmate-only` 配置或 profile。
|
||||
- [ ] 非 AI SoulMate 数据源和模块在该模式下不初始化。
|
||||
- [ ] 管理端对象存储改用同一 provider/兼容接口。
|
||||
- [ ] 清理默认账号、测试数据和固定密码。
|
||||
|
||||
验收:只配置 AI SoulMate 数据库即可启动管理端并完成核心管理功能。
|
||||
|
||||
## P3:可交付单机编排
|
||||
|
||||
- [ ] 编排网关、前后端、管理端、PostgreSQL、Redis 和对象存储。
|
||||
- [ ] 所有镜像固定版本或 digest,服务带健康检查与资源限制。
|
||||
- [ ] 提供初始化、升级、回滚、备份、恢复和日志收集脚本。
|
||||
- [ ] 将持久化目录与容器生命周期分离,定义磁盘容量告警。
|
||||
- [ ] 默认只开放 `22`(受限管理)和 `443`(业务)端口。
|
||||
|
||||
验收:在干净 Ubuntu 22.04 主机上按文档可重复安装、重启和恢复。
|
||||
|
||||
## P4:样机 POC
|
||||
|
||||
- [ ] 激活有线网络并分配固定地址,校准 NTP,验证 DNS 与 Tripo 出站访问。
|
||||
- [ ] 导入脱敏数据库和对象样本,执行数据库迁移。
|
||||
- [ ] 回归登录、模型生成、上传下载、订单/任务和管理端流程。
|
||||
- [ ] 完成一次备份恢复演练和宿主机重启演练。
|
||||
|
||||
验收:核心业务链路跑通,重启无数据丢失,备份能够在空环境恢复。
|
||||
|
||||
## P5:农场工作站接入
|
||||
|
||||
- [ ] 将前端改为生产构建,不使用 Vite 开发服务器。
|
||||
- [ ] 每台设备使用独立身份和可撤销凭证,替换全局客户端令牌。
|
||||
- [ ] 实现任务幂等、断线续传、失败重试和本地队列上限。
|
||||
- [ ] 验证 Moonraker/打印机协议、升级和远程诊断。
|
||||
|
||||
验收:中心短暂断网后任务不重复、不丢失,恢复后状态最终一致。
|
||||
|
||||
## P6:容量测试与客户档位
|
||||
|
||||
- [ ] 先测本地只读、写入、文件和混合负载。
|
||||
- [ ] 再测包含 Tripo 的端到端链路,并剥离外部等待时间。
|
||||
- [ ] 分别测试当前 32GB 和内存升级后的配置。
|
||||
- [ ] 将稳定并发、瓶颈、资源余量和推荐参数写入版本化报告。
|
||||
|
||||
验收:每个客户硬件档位都由可复现的测试数据支撑,并明确适用业务模型与 SLA。
|
||||
|
||||
@@ -0,0 +1,51 @@
|
||||
# POC 样机硬件基线
|
||||
|
||||
基线日期:2026-08-05。数据来自现场命令截图;首次联网部署前应运行 `scripts/collect-host-facts.sh` 生成机器可读基线并复核。
|
||||
|
||||
## 已确认配置
|
||||
|
||||
| 项目 | 当前配置 |
|
||||
| --- | --- |
|
||||
| 设备 | HP ZCentral 4R Workstation |
|
||||
| 操作系统 | Ubuntu 22.04.2 LTS |
|
||||
| 内核/架构 | Linux 5.19.0-32-generic,x86_64 |
|
||||
| CPU | Intel Xeon W-2225,4 核 8 线程,单路,最高约 4.6GHz |
|
||||
| NUMA | 1 个节点 |
|
||||
| 内存 | 32GiB ECC DDR4,单条,标称 3200 MT/s,配置 2933 MT/s |
|
||||
| Swap | 2GiB |
|
||||
| 磁盘 | 单块约 1TB Samsung NVMe,根分区 ext4 |
|
||||
| 可用空间 | 根分区约 882GiB 可用(截图时) |
|
||||
| 网络 | `enp1s0` 与 `eno1` 均为 DOWN,尚未形成网络基线 |
|
||||
|
||||
为避免泄露现场资产信息,本仓库不记录主机序列号、UUID、磁盘序列号、MAC 地址和真实 IP。
|
||||
|
||||
## 适用范围
|
||||
|
||||
这台机器适合:
|
||||
|
||||
- 单机容器化适配与功能回归;
|
||||
- PostgreSQL、Redis、对象存储和业务服务共机时的容量边界测试;
|
||||
- 识别 CPU、内存、磁盘和外部 API 的首个瓶颈;
|
||||
- 为后续不同硬件档位建立对照基线。
|
||||
|
||||
它不代表生产高可用能力:只有一块系统/数据盘,任一磁盘故障都会影响全部服务;4 核 CPU 也会让数据库、对象存储与应用争抢资源。
|
||||
|
||||
## 部署前动作
|
||||
|
||||
1. 接通有线网络,确定固定 IP、DNS、NTP、默认路由和 Tripo 所需出站策略。
|
||||
2. 更新 Ubuntu 22.04 安全补丁,记录 BIOS、内核与 Docker 版本。
|
||||
3. 确认 NVMe SMART/健康状态;生产方案必须设计独立数据盘、备份目标或冗余。
|
||||
4. 将 Swap 调整视为防 OOM 保护而非可用内存,压测报告必须记录 swap in/out。
|
||||
5. 若条件允许,增加同规格 ECC 内存至 64GB,再做一组对照测试;不要把升级后的结论套用到 32GB 配置。
|
||||
6. 压测期间关闭无关桌面应用,固定电源/性能策略,并记录所有参数变化。
|
||||
|
||||
## 当前风险
|
||||
|
||||
| 风险 | 对测试的影响 | 处理方式 |
|
||||
| --- | --- | --- |
|
||||
| 单块 NVMe | 数据库、对象存储、日志相互争抢 I/O | 监测磁盘延迟和利用率;容量报告明确单盘条件 |
|
||||
| 单条 32GB 内存 | 内存容量和带宽可能先成为瓶颈 | 记录内存/Swap;增加 64GB 对照组 |
|
||||
| 4 核 8 线程 | 共机服务容易发生 CPU 争用 | 设置容器资源上限并逐级加压 |
|
||||
| 网卡未启用 | 无法验证局域网和 Tripo 端到端链路 | 部署前先完成网络基线与出站测试 |
|
||||
| 无冗余 | 不适合据此宣称生产可用性 | 将 POC 容量与生产可靠性方案分开 |
|
||||
|
||||
@@ -0,0 +1,29 @@
|
||||
# 安全与发布阻断项
|
||||
|
||||
## P0:凭据轮换
|
||||
|
||||
四个远端仓库的当前检出版本中存在被 Git 跟踪的非示例 .env 或 .env.server。ai-soulmate-server 的 Compose 还包含真实或疑似真实的支付密钥、私钥和固定数据库密码。
|
||||
|
||||
这些值必须视为已泄露:
|
||||
|
||||
1. 先在支付、短信、邮件、数据库、Redis、对象存储、JWT 和外部 AI 服务侧轮换或撤销。
|
||||
2. 从当前源码和后续镜像中移除真实值,只保留变量名与安全示例。
|
||||
3. 检查 Git 历史和镜像层;历史清理需要团队协调,不能在共享仓库上直接强推。
|
||||
4. 用服务器本地 0600 权限的 .env、Docker secrets 或客户密钥系统注入。
|
||||
|
||||
部署仓库不保存任何发现到的真实值。
|
||||
|
||||
## P0:选择发布基线
|
||||
|
||||
ai-soulmate-server 默认分支 develop6 的 3D 集成是 Hunyuan3D/Bailian,不是用户描述的 Tripo;旧交接 SHA 也不是任何远端分支的当前 tip。ai-soulmate-client 的旧交接 SHA 位于 fix/client-login-403,而默认 master 已前进。
|
||||
|
||||
在用户或仓库负责人明确 server/client 的发布分支或 tag 前,不开始业务代码改造,也不生成应用镜像。选择后应创建 local-deployment 分支并把 SHA 写入 inventory。
|
||||
|
||||
## P1:开发配置不能上线
|
||||
|
||||
- server 使用 Uvicorn reload、Vite 开发服务器、源码 bind mount、latest 镜像和对外数据库/Redis 端口。
|
||||
- client 前端仍使用 Vite 开发服务器和源码 bind mount,数据库使用固定密码并对宿主机开放。
|
||||
- admin 同时强依赖四个数据源,尚未验证 AI SoulMate-only 模式。
|
||||
- nginx 绑定现有公网域名、证书路径和固定 upstream,不能直接交付给内网客户。
|
||||
|
||||
应用 Compose 必须等到生产镜像、最小权限、健康检查、资源上限、迁移、备份恢复、端口收敛和凭据注入全部完成。
|
||||
@@ -0,0 +1,38 @@
|
||||
# 源码版本核对
|
||||
|
||||
## 为什么要重新核对
|
||||
|
||||
压缩包中的 inventory/source-baselines.yml 记录了上一台电脑看到的四个提交。只读远端核对后,server 默认分支是 develop6@9a6f7a5,client 默认分支是 master@9ebb082;两者均与旧记录不同。admin 和 nginx 与旧记录一致。另一个本地 ai-soulmate-server 目录指向不同的 Gitee/Spring Boot 代码库,不能与内部 FastAPI 仓库混用。
|
||||
|
||||
因此,当前状态是“交接资料已恢复,源码基线尚未复核”,不是“源码已经可以部署”。
|
||||
|
||||
server 默认分支还使用 Hunyuan3D/Bailian,而用户描述的是 Tripo。必须先由仓库负责人明确 server/client 的发布分支或 tag,再创建本地部署适配分支;默认分支不能自动视为发布版本。
|
||||
|
||||
## 在服务器上复核
|
||||
|
||||
把四个源码仓库放在同一个目录,例如 /srv/ai-soulmate/sources,然后执行:
|
||||
|
||||
~~~bash
|
||||
cd /srv/ai-soulmate-local-deployment
|
||||
SOURCE_ROOT=/srv/ai-soulmate/sources \
|
||||
./scripts/record-source-baseline.sh \
|
||||
> inventory/source-check-$(date +%F-%H%M%S).yml
|
||||
~~~
|
||||
|
||||
提交前检查:
|
||||
|
||||
~~~bash
|
||||
git diff -- inventory/source-check-*.yml
|
||||
git status --short
|
||||
~~~
|
||||
|
||||
报告可以提交,但真实 .env、Token、证书、用户数据和原始日志不能提交。脚本会标记仓库是否跟踪任何非示例 .env 文件;一旦为 true,先审查历史并轮换凭据,再做远端推送。
|
||||
|
||||
## 当前四个仓库的边界
|
||||
|
||||
| 仓库 | 当前行动 |
|
||||
| --- | --- |
|
||||
| ai-soulmate-server | 先完成本地 PostgreSQL、Redis、S3 provider 和生产镜像适配 |
|
||||
| ai-soulmate-client | 后续放到靠近打印机的工作站,先做设备凭证、离线队列和幂等重试 |
|
||||
| mlkj-admin | 只启用 AI SoulMate 数据源和模块,移除无关初始化与默认账号 |
|
||||
| nginx-server | 只参考网关职责,不直接复制公网域名和路由;应用镜像完成后再决定是否使用 |
|
||||
@@ -0,0 +1,26 @@
|
||||
captured_at: "2026-08-05"
|
||||
method: "git ls-remote --symref <repository> HEAD"
|
||||
|
||||
repositories:
|
||||
- name: ai-soulmate-server
|
||||
default_branch: develop6
|
||||
head: "9a6f7a5b336f5b67b1bd699c02280cad76b7332e"
|
||||
handoff_commit: "24769465d3ac2cb840c5266c8c0def2aacb641aa"
|
||||
handoff_commit_is_advertised_branch_tip: false
|
||||
- name: ai-soulmate-client
|
||||
default_branch: master
|
||||
head: "9ebb082b3fe0344437b6676a47fca014acf60da8"
|
||||
handoff_commit: "170847159a787c7b52e3f9f792d72ea67cbd7fa4"
|
||||
handoff_commit_branch: fix/client-login-403
|
||||
- name: mlkj-admin
|
||||
default_branch: main
|
||||
head: "40bd686128010fbbc6c33f6156a7a3cda5336b3a"
|
||||
handoff_commit_matches_head: true
|
||||
- name: nginx-server
|
||||
default_branch: master
|
||||
head: "6f552f4b37505ec055cdd179fe9cc5a6825a4189"
|
||||
handoff_commit_matches_head: true
|
||||
|
||||
notes:
|
||||
- "A default branch is not automatically an approved release branch."
|
||||
- "The server repository contains many feature/fix branch tips and needs an explicit release-baseline decision."
|
||||
@@ -0,0 +1,24 @@
|
||||
captured_at: "2026-08-05"
|
||||
sources:
|
||||
- name: ai-soulmate-server
|
||||
remote: http://39.106.119.229:3000/MLKJ/ai-soulmate-server
|
||||
commit: 24769465d3ac2cb840c5266c8c0def2aacb641aa
|
||||
local_status: clean
|
||||
- name: ai-soulmate-client
|
||||
remote: http://39.106.119.229:3000/MLKJ/ai-soulmate-client
|
||||
commit: 170847159a787c7b52e3f9f792d72ea67cbd7fa4
|
||||
local_status: clean
|
||||
- name: mlkj-admin
|
||||
remote: http://39.106.119.229:3000/MLKJ/mlkj-admin.git
|
||||
commit: 40bd686128010fbbc6c33f6156a7a3cda5336b3a
|
||||
local_status: clean
|
||||
- name: nginx-server
|
||||
remote: http://39.106.119.229:3000/MLKJ/nginx-server
|
||||
commit: 6f552f4b37505ec055cdd179fe9cc5a6825a4189
|
||||
local_status: remote-inspection-only
|
||||
|
||||
notes:
|
||||
- This file records source state only; it does not prove deployment compatibility.
|
||||
- Future releases must add container image digests and database migration versions.
|
||||
- Do not add credentials or access tokens to this inventory.
|
||||
|
||||
@@ -0,0 +1,66 @@
|
||||
captured_at: "2026-08-05"
|
||||
source_workspace: "E:/moling/项目开发/开发"
|
||||
|
||||
observed:
|
||||
- name: ai-soulmate-server-internal
|
||||
path: "../ai-soulmate-server-gitea"
|
||||
remote: "http://39.106.119.229:3000/MLKJ/ai-soulmate-server"
|
||||
branch: develop6
|
||||
commit: "9a6f7a5b336f5b67b1bd699c02280cad76b7332e"
|
||||
local_status: clean
|
||||
deployment_notes:
|
||||
- "The default branch is not an approved release baseline; the repository has many active feature and fix branches."
|
||||
- "The recorded handoff commit is not the tip of any advertised remote branch."
|
||||
- "The checked-out stack uses FastAPI, Vue/Vite, PostgreSQL and Redis in development mode."
|
||||
- "The checked-out storage module implements local filesystem storage only despite mentioning OSS/S3 in its module description."
|
||||
- "The checked-out 3D integration references Hunyuan3D and Alibaba Bailian, not Tripo; confirm the intended release branch."
|
||||
- "A real environment file and payment/private-key material are tracked or embedded in repository files; rotate them before deployment."
|
||||
- name: ai-soulmate-server-other-local-copy
|
||||
path: "../ai-soulmate-server"
|
||||
remote: "https://gitee.com/xppsec/ai-soulmate-server.git"
|
||||
branch: master
|
||||
commit: "e7c3027399ba1c4f3f5327e51111e4705287e1b4"
|
||||
local_status: clean
|
||||
deployment_notes:
|
||||
- "This is a different Spring Boot codebase and must not be mixed with the internal FastAPI repository."
|
||||
- "backend-java/.env is tracked in this workspace."
|
||||
- name: ai-soulmate-client
|
||||
path: "../ai-soulmate-client"
|
||||
remote: "http://39.106.119.229:3000/MLKJ/ai-soulmate-client"
|
||||
branch: master
|
||||
commit: "9ebb082b3fe0344437b6676a47fca014acf60da8"
|
||||
local_status: clean
|
||||
deployment_notes:
|
||||
- "The old handoff commit is the tip of fix/client-login-403, not the default branch."
|
||||
- "The frontend still runs the Vite development server with a source bind mount."
|
||||
- "The compose file exposes PostgreSQL and uses a fixed database password; a real .env is tracked."
|
||||
- name: mlkj-admin
|
||||
path: "../mlkj-admin"
|
||||
remote: "http://39.106.119.229:3000/MLKJ/mlkj-admin"
|
||||
branch: main
|
||||
commit: "40bd686128010fbbc6c33f6156a7a3cda5336b3a"
|
||||
local_status: clean
|
||||
deployment_notes:
|
||||
- "The application config requires admin, ai_soulmate, official_site and bcjh datasources."
|
||||
- "There is no verified AI SoulMate-only profile."
|
||||
- "Real .env and .env.server files are tracked."
|
||||
- name: nginx-server
|
||||
path: "../nginx-server"
|
||||
remote: "http://39.106.119.229:3000/MLKJ/nginx-server"
|
||||
branch: master
|
||||
commit: "6f552f4b37505ec055cdd179fe9cc5a6825a4189"
|
||||
local_status: clean
|
||||
deployment_notes:
|
||||
- "The routes are bound to current public domains, certificate paths and upstream names."
|
||||
- "Real .env and .env.server files are tracked."
|
||||
- "Reuse the gateway responsibility, not this environment-specific configuration."
|
||||
|
||||
recorded_handoff_baseline:
|
||||
ai_soulmate_server: "24769465d3ac2cb840c5266c8c0def2aacb641aa"
|
||||
ai_soulmate_client: "170847159a787c7b52e3f9f792d72ea67cbd7fa4"
|
||||
mlkj_admin: "40bd686128010fbbc6c33f6156a7a3cda5336b3a"
|
||||
nginx_server: "6f552f4b37505ec055cdd179fe9cc5a6825a4189"
|
||||
|
||||
notes:
|
||||
- "Do not deploy or modify business code until the server and client release branches are explicitly selected."
|
||||
- "Do not copy tracked environment files into the deployment repository."
|
||||
@@ -0,0 +1 @@
|
||||
|
||||
@@ -0,0 +1 @@
|
||||
|
||||
Executable
+58
@@ -0,0 +1,58 @@
|
||||
#!/usr/bin/env bash
|
||||
set -Eeuo pipefail
|
||||
|
||||
if [[ "${EUID}" -eq 0 ]]; then
|
||||
sudo_command=()
|
||||
else
|
||||
sudo_command=(sudo)
|
||||
fi
|
||||
|
||||
if [[ ! -r /etc/os-release ]]; then
|
||||
printf '/etc/os-release is unavailable.\n' >&2
|
||||
exit 1
|
||||
fi
|
||||
|
||||
source /etc/os-release
|
||||
if [[ "${ID:-}" != "ubuntu" || "${VERSION_ID:-}" != "22.04" ]]; then
|
||||
printf 'This script requires Ubuntu 22.04; found %s %s.\n' "${ID:-unknown}" "${VERSION_ID:-unknown}" >&2
|
||||
exit 1
|
||||
fi
|
||||
|
||||
timezone_name="${TIMEZONE_NAME:-Asia/Shanghai}"
|
||||
deploy_user="${DEPLOY_USER:-${SUDO_USER:-}}"
|
||||
|
||||
"${sudo_command[@]}" timedatectl set-timezone "$timezone_name"
|
||||
"${sudo_command[@]}" apt-get update
|
||||
"${sudo_command[@]}" apt-get install -y ca-certificates curl gnupg git jq rsync tmux htop sysstat iotop nvme-cli smartmontools unzip
|
||||
|
||||
if [[ "${RUN_UPGRADE:-0}" == "1" ]]; then
|
||||
"${sudo_command[@]}" apt-get upgrade -y
|
||||
fi
|
||||
|
||||
"${sudo_command[@]}" install -m 0755 -d /etc/apt/keyrings
|
||||
curl -fsSL https://download.docker.com/linux/ubuntu/gpg \
|
||||
| "${sudo_command[@]}" gpg --dearmor --yes -o /etc/apt/keyrings/docker.gpg
|
||||
"${sudo_command[@]}" chmod a+r /etc/apt/keyrings/docker.gpg
|
||||
|
||||
architecture="$(dpkg --print-architecture)"
|
||||
codename="${VERSION_CODENAME}"
|
||||
printf 'deb [arch=%s signed-by=/etc/apt/keyrings/docker.gpg] https://download.docker.com/linux/ubuntu %s stable\n' \
|
||||
"$architecture" "$codename" \
|
||||
| "${sudo_command[@]}" tee /etc/apt/sources.list.d/docker.list >/dev/null
|
||||
|
||||
"${sudo_command[@]}" apt-get update
|
||||
"${sudo_command[@]}" apt-get install -y docker-ce docker-ce-cli containerd.io docker-buildx-plugin docker-compose-plugin
|
||||
"${sudo_command[@]}" systemctl enable --now docker
|
||||
|
||||
if [[ -n "$deploy_user" && "$deploy_user" != "root" ]]; then
|
||||
"${sudo_command[@]}" usermod -aG docker "$deploy_user"
|
||||
printf 'Added %s to the docker group; start a new SSH session before using docker without sudo.\n' "$deploy_user"
|
||||
fi
|
||||
|
||||
"${sudo_command[@]}" docker version
|
||||
"${sudo_command[@]}" docker compose version
|
||||
"${sudo_command[@]}" systemctl is-active docker
|
||||
|
||||
if [[ -e /var/run/reboot-required ]]; then
|
||||
printf 'A reboot is required. Keep the current SSH session until key login and the reserved address are rechecked.\n'
|
||||
fi
|
||||
Executable
+36
@@ -0,0 +1,36 @@
|
||||
#!/usr/bin/env bash
|
||||
set -Eeuo pipefail
|
||||
|
||||
section() {
|
||||
printf '\n## %s\n' "$1"
|
||||
}
|
||||
|
||||
section "Snapshot"
|
||||
printf 'captured_at=%s\n' "$(date --iso-8601=seconds)"
|
||||
uptime
|
||||
|
||||
section "Memory"
|
||||
free -h
|
||||
if command -v vmstat >/dev/null 2>&1; then
|
||||
vmstat 1 5
|
||||
fi
|
||||
|
||||
section "Storage"
|
||||
df --human-readable --print-type
|
||||
if command -v iostat >/dev/null 2>&1; then
|
||||
iostat -xz 1 3
|
||||
fi
|
||||
|
||||
section "Network summary"
|
||||
if command -v ss >/dev/null 2>&1; then
|
||||
ss --summary
|
||||
fi
|
||||
|
||||
section "Containers"
|
||||
if command -v docker >/dev/null 2>&1; then
|
||||
docker ps --format 'table {{.Names}}\t{{.Image}}\t{{.Status}}\t{{.Ports}}'
|
||||
docker stats --no-stream --format 'table {{.Name}}\t{{.CPUPerc}}\t{{.MemUsage}}\t{{.NetIO}}\t{{.BlockIO}}\t{{.PIDs}}'
|
||||
else
|
||||
printf '%s\n' 'docker is not installed'
|
||||
fi
|
||||
|
||||
Executable
+67
@@ -0,0 +1,67 @@
|
||||
#!/usr/bin/env bash
|
||||
set -Eeuo pipefail
|
||||
|
||||
section() {
|
||||
printf '\n## %s\n' "$1"
|
||||
}
|
||||
|
||||
run_optional() {
|
||||
local command_name="$1"
|
||||
shift
|
||||
|
||||
if command -v "$command_name" >/dev/null 2>&1; then
|
||||
"$command_name" "$@" 2>&1 || true
|
||||
else
|
||||
printf '%s is not installed\n' "$command_name"
|
||||
fi
|
||||
}
|
||||
|
||||
section "Collection"
|
||||
printf 'collected_at=%s\n' "$(date --iso-8601=seconds)"
|
||||
printf 'collector_version=1\n'
|
||||
|
||||
section "Operating system"
|
||||
if [[ -r /etc/os-release ]]; then
|
||||
sed -n -E '/^(NAME|VERSION|VERSION_ID|VERSION_CODENAME|PRETTY_NAME)=/p' /etc/os-release
|
||||
fi
|
||||
run_optional uname -a
|
||||
run_optional timedatectl status
|
||||
|
||||
section "CPU"
|
||||
run_optional lscpu
|
||||
|
||||
section "Memory"
|
||||
run_optional free -h
|
||||
run_optional swapon --show --bytes
|
||||
if command -v sudo >/dev/null 2>&1 && sudo -n true 2>/dev/null && command -v dmidecode >/dev/null 2>&1; then
|
||||
sudo -n dmidecode --type memory 2>/dev/null \
|
||||
| sed -n -E '/^[[:space:]]*(Error Correction Type|Size|Form Factor|Locator|Bank Locator|Type|Speed|Configured Memory Speed):/p'
|
||||
else
|
||||
printf 'dmidecode skipped: passwordless sudo or dmidecode is unavailable\n'
|
||||
fi
|
||||
|
||||
section "Block devices"
|
||||
run_optional lsblk --output NAME,SIZE,TYPE,FSTYPE,MOUNTPOINTS,MODEL
|
||||
run_optional df --human-readable --print-type
|
||||
|
||||
section "Network link state"
|
||||
for interface_path in /sys/class/net/*; do
|
||||
[[ -e "$interface_path" ]] || continue
|
||||
interface_name="${interface_path##*/}"
|
||||
interface_state="$(cat "$interface_path/operstate" 2>/dev/null || printf 'unknown')"
|
||||
printf '%s state=%s\n' "$interface_name" "$interface_state"
|
||||
if command -v ethtool >/dev/null 2>&1 && [[ "$interface_name" != "lo" ]]; then
|
||||
ethtool "$interface_name" 2>/dev/null \
|
||||
| sed -n -E '/^[[:space:]]*(Speed|Duplex|Auto-negotiation|Link detected):/p' || true
|
||||
fi
|
||||
done
|
||||
|
||||
section "Container runtime"
|
||||
run_optional docker version
|
||||
if command -v docker >/dev/null 2>&1; then
|
||||
docker compose version 2>&1 || true
|
||||
fi
|
||||
|
||||
section "Security notice"
|
||||
printf '%s\n' 'This collector intentionally omits IP addresses, MAC addresses, serial numbers and machine UUIDs.'
|
||||
|
||||
Executable
+54
@@ -0,0 +1,54 @@
|
||||
#!/usr/bin/env bash
|
||||
set -Eeuo pipefail
|
||||
|
||||
target_path="${1:-compose/.env}"
|
||||
force_write="${FORCE_WRITE:-0}"
|
||||
|
||||
if [[ -e "$target_path" && "$force_write" != "1" ]]; then
|
||||
printf 'Refusing to overwrite %s; set FORCE_WRITE=1 to replace it.\n' "$target_path" >&2
|
||||
exit 1
|
||||
fi
|
||||
|
||||
if ! command -v openssl >/dev/null 2>&1; then
|
||||
printf 'openssl is required to generate local secrets.\n' >&2
|
||||
exit 1
|
||||
fi
|
||||
|
||||
target_directory="$(dirname -- "$target_path")"
|
||||
mkdir -p -- "$target_directory"
|
||||
umask 077
|
||||
|
||||
random_secret() {
|
||||
openssl rand -hex 32
|
||||
}
|
||||
|
||||
cat > "$target_path" <<EOF
|
||||
COMPOSE_PROJECT_NAME=ai-soulmate-infra
|
||||
DOCKER_NETWORK_NAME=ai-soulmate-internal
|
||||
TZ=Asia/Shanghai
|
||||
|
||||
POSTGRES_IMAGE=postgres:16-alpine
|
||||
POSTGRES_DB=ai_soulmate
|
||||
POSTGRES_USER=ai_soulmate
|
||||
POSTGRES_PASSWORD=$(random_secret)
|
||||
POSTGRES_HOST_PORT=5432
|
||||
POSTGRES_VOLUME_NAME=ai-soulmate-postgres-data
|
||||
|
||||
REDIS_IMAGE=redis:7-alpine
|
||||
REDIS_PASSWORD=$(random_secret)
|
||||
REDIS_HOST_PORT=6379
|
||||
REDIS_VOLUME_NAME=ai-soulmate-redis-data
|
||||
|
||||
MINIO_IMAGE=minio/minio:latest
|
||||
MINIO_MC_IMAGE=minio/mc:latest
|
||||
MINIO_ROOT_USER=ai_soulmate_admin
|
||||
MINIO_ROOT_PASSWORD=$(random_secret)
|
||||
MINIO_BUCKET=ai-soulmate
|
||||
MINIO_API_HOST_PORT=9000
|
||||
MINIO_CONSOLE_HOST_PORT=9001
|
||||
MINIO_VOLUME_NAME=ai-soulmate-minio-data
|
||||
EOF
|
||||
|
||||
chmod 600 "$target_path"
|
||||
printf 'Created %s with generated secrets.\n' "$target_path"
|
||||
printf 'Replace mutable image tags with verified tags or digests before a release.\n'
|
||||
Executable
+48
@@ -0,0 +1,48 @@
|
||||
#!/usr/bin/env bash
|
||||
set -Eeuo pipefail
|
||||
|
||||
source_root="${SOURCE_ROOT:-../sources}"
|
||||
captured_at="$(date --iso-8601=seconds)"
|
||||
|
||||
printf 'captured_at: "%s"\n' "$captured_at"
|
||||
printf 'source_root: "%s"\n' "$source_root"
|
||||
printf 'sources:\n'
|
||||
|
||||
repo_names=(ai-soulmate-server ai-soulmate-client mlkj-admin nginx-server)
|
||||
for repo_name in "${repo_names[@]}"; do
|
||||
repo_path="$source_root/$repo_name"
|
||||
printf ' - name: "%s"\n' "$repo_name"
|
||||
printf ' path: "%s"\n' "$repo_path"
|
||||
if [[ ! -d "$repo_path/.git" ]]; then
|
||||
printf ' available: false\n'
|
||||
printf ' reason: "git repository not found"\n'
|
||||
continue
|
||||
fi
|
||||
|
||||
commit_id="$(git -C "$repo_path" rev-parse HEAD)"
|
||||
branch_name="$(git -C "$repo_path" branch --show-current)"
|
||||
remote_url="$(git -C "$repo_path" remote get-url origin 2>/dev/null || true)"
|
||||
sanitized_remote="$(printf '%s' "$remote_url" | sed -E 's#(https?://)[^/@]+@#\1#')"
|
||||
if [[ -n "$(git -C "$repo_path" status --porcelain)" ]]; then
|
||||
local_status=dirty
|
||||
else
|
||||
local_status=clean
|
||||
fi
|
||||
tracked_env_files="$(git -C "$repo_path" ls-files | grep -E '(^|/)\.env($|\.)' | grep -vE '\.example$' || true)"
|
||||
if [[ -n "$tracked_env_files" ]]; then
|
||||
tracked_env=true
|
||||
else
|
||||
tracked_env=false
|
||||
fi
|
||||
|
||||
printf ' available: true\n'
|
||||
printf ' branch: "%s"\n' "$branch_name"
|
||||
printf ' commit: "%s"\n' "$commit_id"
|
||||
printf ' local_status: "%s"\n' "$local_status"
|
||||
printf ' origin: "%s"\n' "$sanitized_remote"
|
||||
printf ' tracked_env_files: %s\n' "$tracked_env"
|
||||
done
|
||||
|
||||
printf 'notes:\n'
|
||||
printf ' - "This report records the checked-out state only; it does not validate deployment compatibility."\n'
|
||||
printf ' - "Review tracked environment files and rotate any credentials before deployment."\n'
|
||||
Executable
+27
@@ -0,0 +1,27 @@
|
||||
#!/usr/bin/env bash
|
||||
set -Eeuo pipefail
|
||||
|
||||
env_file="${1:-compose/.env}"
|
||||
compose_file="${COMPOSE_FILE:-compose/docker-compose.infra.yml}"
|
||||
|
||||
if [[ ! -r "$env_file" ]]; then
|
||||
printf 'Environment file not found: %s\n' "$env_file" >&2
|
||||
exit 1
|
||||
fi
|
||||
|
||||
compose_command=(docker compose --env-file "$env_file" -f "$compose_file")
|
||||
|
||||
"${compose_command[@]}" config --quiet
|
||||
"${compose_command[@]}" exec -T postgres sh -ec 'pg_isready -U "$POSTGRES_USER" -d "$POSTGRES_DB"'
|
||||
"${compose_command[@]}" exec -T redis sh -ec 'redis-cli -a "$REDIS_PASSWORD" ping 2>/dev/null | grep -q PONG'
|
||||
"${compose_command[@]}" exec -T minio sh -ec 'curl -fsS http://127.0.0.1:9000/minio/health/live >/dev/null'
|
||||
|
||||
minio_init_status="$("${compose_command[@]}" ps --all --format json minio-init)"
|
||||
if ! printf '%s' "$minio_init_status" \
|
||||
| jq -e 'if type == "array" then length == 1 and .[0].ExitCode == 0 else .ExitCode == 0 end' >/dev/null; then
|
||||
printf 'MinIO bucket initialization did not complete successfully.\n' >&2
|
||||
exit 1
|
||||
fi
|
||||
|
||||
"${compose_command[@]}" ps
|
||||
printf 'Infrastructure checks passed.\n'
|
||||
@@ -0,0 +1,64 @@
|
||||
---
|
||||
name: ai-soulmate-local-deployment
|
||||
description: Guide cross-computer continuation of the AI SoulMate Ubuntu private-deployment work, including SSH access, hardware baselines, Git deployment archival, Docker prerequisites, service-boundary decisions, and benchmark preparation. Use when Codex needs to resume, document, or operationalize AI SoulMate local deployment on an Ubuntu 22.04 adaptation server.
|
||||
---
|
||||
|
||||
# AI SoulMate Local Deployment
|
||||
|
||||
Use this skill to resume the AI SoulMate private-deployment adaptation without treating an unfinished POC as production-ready. Keep the deployment/orchestration repository separate from the three business-source repositories.
|
||||
|
||||
## First actions
|
||||
|
||||
1. Locate the deployment repository and read `docs/CROSS-COMPUTER-HANDOFF.md`, `docs/IMPLEMENTATION-ROADMAP.md`, `docs/ARCHITECTURE.md`, and `inventory/source-baselines.yml` when they are present.
|
||||
2. Check whether the deployment repository exists, whether it has a Git history, and whether a remote is configured. Do not assume that work from another computer is present.
|
||||
3. Inspect each source repository with `git status --short` and `git rev-parse HEAD`. Preserve unrelated working-tree changes.
|
||||
4. Record the current host facts before changing packages, network settings, storage, or container limits.
|
||||
|
||||
## Repository boundary
|
||||
|
||||
- Store Compose files, environment examples, version pins, migrations manifest, backup procedures, host-fact collectors, benchmark metadata, and reports in the deployment repository.
|
||||
- Keep Python/Vue platform code in `ai-soulmate-server`, farm-side code in `ai-soulmate-client`, and management code in `mlkj-admin`.
|
||||
- Record source SHAs and image digests; do not copy source trees into the deployment repository or use submodules as the release mechanism.
|
||||
- Do not add passwords, API keys, certificates, customer data, machine UUIDs, MAC addresses, or raw sensitive logs to Git.
|
||||
|
||||
## SSH and network workflow
|
||||
|
||||
- Connect to a local console first. Confirm the wired interface has a routable address with `nmcli device status` and `ip -br addr`.
|
||||
- Treat `enp1s0` and `eno1` as examples, not assumptions; connect only the interface with the cable and expected network profile.
|
||||
- Use a DHCP reservation or an approved static address. A private address such as `192.168.3.102` is reachable from another computer only when routing, VLAN policy, and Wi-Fi isolation permit it.
|
||||
- Install `openssh-server`, enable `ssh`, and validate with `sudo sshd -t` and `ss -lntp`.
|
||||
- Establish Ed25519 key login from the operator computer. Keep the password-login session open until a second key-based session is proven.
|
||||
- Never expose SSH, databases, Redis, or object-storage administration ports to the public Internet. Use an approved VPN for cross-site access.
|
||||
|
||||
## Host preparation
|
||||
|
||||
1. Save a sanitized host baseline: Ubuntu version, kernel, CPU, memory, swap, block layout, filesystem capacity, link state, Docker version, and time synchronization. Omit IP, MAC, serial, and UUID unless the user explicitly requests an internal inventory.
|
||||
2. Set the approved timezone and verify NTP.
|
||||
3. Apply Ubuntu security updates and install only necessary tools (`git`, `curl`, `ca-certificates`, `gnupg`, `jq`, `rsync`, `tmux`, monitoring and storage tools).
|
||||
4. Install Docker Engine, Buildx, and Compose plugin from Docker's official Ubuntu repository. Do not use an unreviewed curl-to-shell installer.
|
||||
5. Reboot only after recording whether a reboot is required, then recheck the reserved IP, SSH, Docker, disk space, and memory.
|
||||
|
||||
## Deployment gates
|
||||
|
||||
Do not publish a runnable production Compose until these gates pass:
|
||||
|
||||
- P1: platform and admin Alibaba OSS calls are behind a storage provider interface; a local S3-compatible provider passes presigned upload/download, CORS, multipart, metadata, permission, backup, and restore tests.
|
||||
- P2: `mlkj-admin` has an AI SoulMate-only mode that does not initialize unrelated databases; default accounts and fixed passwords are removed from delivery data.
|
||||
- P3: images are immutable and version-pinned; no source bind mounts; services have health checks, resource limits, persistent volumes, migrations, backups, restore, and controlled ports.
|
||||
- P4: core workflow works with Tripo outbound access, host reboot, and backup recovery; Tripo remains explicitly marked as an external dependency.
|
||||
- P5: farm workstation has device-specific credentials, idempotent tasks, retry/offline behavior, and printer/Moonraker validation.
|
||||
|
||||
The current public-domain `nginx-server` configuration is not a private deployment artifact. Reuse the gateway responsibility, but create environment-neutral routes for platform frontend, backend, admin, and later farm endpoints; use a customer gateway when one already exists.
|
||||
|
||||
## Benchmark workflow
|
||||
|
||||
- Separate local-only capacity from Tripo end-to-end experience. Report Tripo submit, polling, download, rate-limit, and external queue time independently.
|
||||
- Use fixed datasets and request mixes. Ramp concurrency gradually (`1, 2, 5, 10, 15, 20...`), warm up, hold a 15–30 minute steady period, cool down, and repeat each point at least three times.
|
||||
- Capture throughput, success/error rate, p50/p95/p99, queue depth, CPU/load, memory and swap, disk latency/utilization, network, container restarts/OOM, database pool/slow queries/locks, Redis latency/eviction, object-storage latency, and Tripo timing/errors.
|
||||
- Stop on OOM, sustained swapping, repeated restarts, data errors, growing queue, exhausted pools, or sustained SLA/error/latency failure. Do not turn one failed point into a customer concurrency promise.
|
||||
- Archive sanitized metadata and summary reports; keep large raw logs outside Git unless explicitly approved.
|
||||
|
||||
## Handoff output
|
||||
|
||||
At the end of each session, update the handoff document with completed commands, exact versions, unresolved blockers, and the next safe action. If the deployment repository is new, show its Git status and whether the first commit and remote push were completed; never claim archival until a commit exists.
|
||||
|
||||
@@ -0,0 +1,5 @@
|
||||
interface:
|
||||
display_name: "AI SoulMate 本地部署接管"
|
||||
short_description: "跨电脑继续 Ubuntu 本地部署、SSH、Git 留档与压测准备"
|
||||
default_prompt: "使用 $ai-soulmate-local-deployment 接管 AI SoulMate 的 Ubuntu 本地部署、SSH 配置、版本留档和压测准备。"
|
||||
|
||||
Reference in new issue
Block a user