ov-server-operate

OpenViking 服务运维

本指南提供在生产环境中部署、管理和维护 OpenViking 服务的标准操作流程。

目录

• 服务配置
• 使用 uv 搭建环境
• 使用 nohup 启动服务
• 停止服务
• 数据清理流程
• 验证与故障排查

服务配置

默认路径与结构

OpenViking 在 ~/.openviking/ 下使用以下标准目录结构：

~/.openviking/
├── ov.conf             # Server configuration (required)
├── ovcli.conf          # CLI client configuration
├── ov-venv/            # Virtual environment (created by uv)
├── log/                # Server log directory
│   ├── openviking-server.log   # server stdout log
│   └── openviking.log          # server log
└── data/               # Workspace data (configured in ov.conf)
    ├── ...
    └── ...

配置文件

1. 服务配置（~/.openviking/ov.conf）

创建配置文件，至少包含以下配置。 注意 1：将 api-key 替换为你自己的 api-key。如果你没有，请让用户获取一个（按照火山引擎方舟平台指南操作）。 注意 2：将 root_api_key 替换为你自己的 root-api-key。请让用户设置一个——它将在 CLI 连接服务时用于身份验证。

{
  "server": {
    "host": "0.0.0.0",
    "port": 1933,
    "root_api_key": "your-root-api-key"
  },
  "storage": {
    "workspace": "~/.openviking/data/"
  },
  "parsers": {
    "code": {
      "gitlab_domains": ["code.byted.org"],
      "azure_devops_domains": ["ssh.dev.azure.com", "vs-ssh.visualstudio.com"]
    }
  },
  "embedding": {
    "dense": {
        "model": "doubao-embedding-vision-251215",
        "api_key": "your-volcengine-api-key",
        "api_base": "https://ark.cn-beijing.volces.com/api/v3",
        "dimension": 1024,
        "input": "multimodal",
        "provider": "volcengine"
    }
  },
  "vlm": {
    "model": "doubao-seed-1-8-251228",
    "api_key": "your-volcengine-api-key",
    "api_base": "https://ark.cn-beijing.volces.com/api/v3",
    "temperature": 0.0,
    "max_retries": 2,
    "provider": "volcengine",
    "thinking": false
  },
  "log": {
    "level": "INFO",
    "output": "file",
    "rotation": true,
    "rotation_days": 3,
    "rotation_interval": "midnight"
  }
}

2. CLI 配置（~/.openviking/ovcli.conf）

用于从 localhost 发起的客户端连接：

{
  "url": "http://localhost:1933",
  "api_key": "your-root-api-key"
}

对于远程连接，将 url 设置为远程服务地址（例如服务 EIP）。

使用 uv 搭建环境

步骤 1：安装 uv（如果尚未安装）

# macOS/Linux
curl -LsSf https://astral.sh/uv/install.sh | sh

# Verify installation
uv --version

步骤 2：创建虚拟环境

在 ~/.openviking/ov-venv 创建专用虚拟环境：

# Create venv with Python 3.10+
cd ~/.openviking
uv venv --python 3.12 ov-venv

步骤 3：激活并安装 OpenViking

# Activate the virtual environment
source ~/.openviking/ov-venv/bin/activate

# Install or upgrade to latest openviking
uv pip install --upgrade openviking --force-reinstall

# Verify installation
which openviking-server
openviking-server --version
openviking-server --help

步骤 4：创建日志目录

mkdir -p ~/.openviking/log

使用 nohup 启动服务

标准启动流程

# 1. Activate the virtual environment
source ~/.openviking/ov-venv/bin/activate

# 2. Ensure log directory exists
mkdir -p ~/.openviking/log

# 3. Start server with nohup
nohup openviking-server \
    > ~/.openviking/log/openviking-server.log 2>&1 &

# 4. Save PID for later reference
echo $! > ~/.openviking/server.pid

# 5. Verify startup after 10 secs
sleep 10
curl -s http://localhost:1933/health

验证服务正在运行

# Method 1: Check health endpoint
curl http://localhost:1933/health
# Expected: {"status": "ok"}

# Method 2: Check readiness (includes storage checks)
curl http://localhost:1933/ready

# Method 3: Check process
ps aux | grep openviking-server | grep -v grep

# Method 4: Check log output
tail -10 ~/.openviking/log/openviking-server.log
tail -50 ~/.openviking/log/openviking.log

停止服务

优雅停止流程

# 1. Find the server process
ps aux | grep openviking-server | grep -v grep

# 2. Send SIGTERM for graceful shutdown
# Option A: Using saved PID
if [ -f ~/.openviking/server.pid ]; then
    kill $(cat ~/.openviking/server.pid)
    rm ~/.openviking/server.pid
fi

# Option B: Using pgrep
pkill -f openviking-server

# 3. Wait for process to stop
sleep 3

# 4. Verify it stopped
ps aux | grep openviking-server | grep -v grep || echo "Server stopped successfully"

# 5. If still running, force kill
if pgrep -f openviking-server > /dev/null; then
    echo "Force killing server..."
    pkill -9 -f openviking-server
fi

数据清理流程

何时使用此流程

在以下场景中执行完整数据清理：

1. 版本升级且数据格式不兼容
2. 数据损坏或不一致
3. 需要重置为全新状态
4. 回收存储空间

标准清理工作流

关键：删除数据前务必先备份

# ==========================================
# STEP 1: STOP THE SERVER FIRST
# ==========================================
echo "Step 1: Stopping OpenViking Server..."
if pgrep -f openviking-server > /dev/null; then
    pkill -f openviking-server
    sleep 3
    if pgrep -f openviking-server > /dev/null; then
        pkill -9 -f openviking-server
        sleep 1
    fi
fi

# Verify server is stopped
if pgrep -f openviking-server > /dev/null; then
    echo "ERROR: Server still running! Cannot proceed."
    exit 1
fi
echo "✓ Server stopped"

# ==========================================
# STEP 2: CREATE BACKUP (REQUIRED)
# ==========================================
echo ""
echo "Step 2: Creating backup..."
BACKUP_DATE=$(date +%Y%m%d_%H%M%S)
BACKUP_DIR=~/.openviking/backup_${BACKUP_DATE}

mkdir -p ${BACKUP_DIR}

# Backup config files
cp ~/.openviking/ov.conf ${BACKUP_DIR}/ 2>/dev/null || true
cp ~/.openviking/ovcli.conf ${BACKUP_DIR}/ 2>/dev/null || true

# Backup workspace (if exists)
WORKSPACE=$(python3 -c '
import json
import os
config_path = os.path.expanduser("~/.openviking/ov.conf")
if os.path.exists(config_path):
    with open(config_path) as f:
        cfg = json.load(f)
        ws = cfg.get("storage", {}).get("workspace", "./data")
        print(os.path.expanduser(ws))
' 2>/dev/null || echo "~/.openviking/data")

if [ -d "${WORKSPACE}" ]; then
    echo "Backing up workspace: ${WORKSPACE}"
    tar -czf ${BACKUP_DIR}/workspace_backup.tar.gz -C $(dirname ${WORKSPACE}) $(basename ${WORKSPACE})
fi

# Backup log
if [ -d ~/.openviking/log ]; then
    cp -r ~/.openviking/log ${BACKUP_DIR}/ 2>/dev/null || true
fi

echo "✓ Backup created at: ${BACKUP_DIR}"
ls -lh ${BACKUP_DIR}/

# ==========================================
# STEP 3: CONFIRM DELETION
# ==========================================
echo ""
echo "=========================================="
echo "WARNING: ABOUT TO DELETE ALL DATA!"
echo "=========================================="
echo "Workspace to delete: ${WORKSPACE}"
echo "Backup location: ${BACKUP_DIR}"
echo ""
read -p "Type 'DELETE' to confirm data removal: " CONFIRM

if [ "${CONFIRM}" != "DELETE" ]; then
    echo "Cleanup cancelled. Backup preserved at ${BACKUP_DIR}"
    exit 0
fi

# ==========================================
# STEP 4: DELETE DATA
# ==========================================
echo ""
echo "Step 4: Deleting data..."

# Delete workspace
if [ -d "${WORKSPACE}" ]; then
    echo "Deleting workspace: ${WORKSPACE}"
    rm -rf "${WORKSPACE}"
fi

# Optional: Delete old log (uncomment if needed)
# echo "Deleting old log..."
# rm -rf ~/.openviking/log/*

# Cleanup any temporary files
rm -f ~/.openviking/server.pid

echo "✓ Data deleted successfully"

# ==========================================
# STEP 5: COMPLETION
# ==========================================
echo ""
echo "=========================================="
echo "Cleanup Complete!"
echo "=========================================="
echo "Backup preserved at: ${BACKUP_DIR}"
echo ""
echo "Next steps:"
echo "1. Reconfigure ov.conf if needed"
echo "2. Start the server: openviking-server"
echo "3. Verify with: curl http://localhost:1933/health"
echo ""
echo "To restore from backup:"
echo "  tar -xzf ${BACKUP_DIR}/workspace_backup.tar.gz -C $(dirname ${WORKSPACE})"

快速清理（仅用于开发）

# WARNING: Only use in development!
# No backup created - data loss guaranteed!

# 1. Stop server
pkill -f openviking-server
sleep 2
pkill -9 -f openviking-server 2>/dev/null || true

# 2. Delete workspace (adjust path as needed)
rm -rf ~/.openviking/data

# 3. Cleanup PID and temp files
rm -f ~/.openviking/server.pid

echo "Quick cleanup complete"

验证与故障排查

健康检查验证

# Basic health check (always available)
curl http://localhost:1933/health
# Expected: {"status": "ok"}

# Readiness check (verifies all components)
curl http://localhost:1933/ready
# Expected: {"status": "ready", "checks": {"agfs": "ok", "vectordb": "ok", "api_key_manager": "ok"}}

# System status via CLI (~/.openviking/ovcli.conf should be configured)
ov status

常见问题与解决方案

问题：服务无法启动

检查：

# 1. Check if port is in use
lsof -i :1933
netstat -tulpn | grep 1933

# 2. Check log for errors
tail -10 ~/.openviking/log/openviking-server.log
tail -100 ~/.openviking/log/openviking.log


# 3. Verify config file is valid JSON
python3 -c 'import json, os; json.load(open(os.path.expanduser("~/.openviking/ov.conf"))); print("Config is valid")'

# 4. Verify virtual environment
source ~/.openviking/ov-venv/bin/activate
which openviking-server
pip list | grep openviking

解决方案：

# If port conflict: kill the process or use different port
lsof -ti :1933 | xargs kill -9 2>/dev/null || true

# Or start on different port
nohup openviking-server --port 1934 > ~/.openviking/log/openviking-server.log 2>&1 &

问题：API Key 错误

检查：

# Verify API keys in config
python3 -c '
import json, os
cfg = json.load(open(os.path.expanduser("~/.openviking/ov.conf")))
print("Embedding provider:", cfg.get("embedding", {}).get("dense", {}).get("provider"))
print("VLM provider:", cfg.get("vlm", {}).get("provider"))
print("API keys set:", bool(cfg.get("embedding", {}).get("dense", {}).get("api_key")), bool(cfg.get("vlm", {}).get("api_key")))
'

解决方案： 验证 API key 是否正确且具有所需权限。检查与模型提供商端点的网络连通性。确保 API key 未过期。

前提条件

• 已安装 Python 3.10+
• 可使用 uv 包管理器
• 有足够的磁盘空间用于工作区和日志
• 已配置用于 embedding 和 VLM 模型的 API key
• 可访问模型提供商的网络（如果使用云端模型）